Comments
Clive Robinson • June 9, 2022 9:13 PM
@ vas pup, ALL,
Re : Red Flag laws,
I am very much against them.
To see why you first have to remove the emotive wish thinking and political issues behind guns.
Red Flag laws are “pre-thought crime prevention” and as such make less sense than using the last digit of a mobile phone number as a predictor of a persons mental state.
No doubt within the next decade much scorn will get poured on the idea of being able to predict a persons future actions as it has been in less contentious areas such as job recruitment.
However certain people will not accept the fact that with only very very few exceptions can you predict how someone will behave in the future.
We are as has oft been said,
“The product of our experiences”
And we have little or no ability to predict our future experiences even fairly close-in in time. Think back to the last time something went wrong like you got a paper-cut, stubbed your toe, bumped into a person in a crowd, sipped from a too hot cup of coffee, or you burnt/scalded yourself in the kitchen. All are fairly easily avoidable “accidents” thus the more annoying because of it. Each and every one is like a billiard ball hitting another it changes our direction in some way. That is every such event changes us, we do not know when or by how much, only that we are like water molecules in hot water we move around in all directions due to such collisions. But seen from a statistical point of view over millions of molecules we know that those molecules with more thermal / vibrational energy will rise above those with less. So whilst on mass you can say where on average a molecule will be on the verticle gradient, you’ve no idea, where in the cross sectional area the molecule will be. We accept “Brownian Motion” in water, but realy don’t consider it with regards to our lives and personalities.
But there are always those who believe in both “pre-ordained” and “free-will” yet do not see the contradiction that arises. Especially when it comes to “what-aboutism” and “wish-fulfillment” predicated on regret and loss.
Such people are cognitively biased into believing we can make such predictions, which is very dangerous. Because of the next step of,
“If humans can not make such predictions then surely AI can”
When such thinking comes into play disaster and excuses are sure to follow, and proponents will keep “doubling down” on what is in reality a proces that at the fine grained level is “random”. Which then as it becomes less fine grained will be chaotic but pull out far enough gets some statistical properties. Thus it is just another form of “gambling” like a roulette wheel, the individual chance that a ball will drop in the green(s) is random, but over enough spins gives a reliable “house advantage”.
Thus AI will get used and as that requires no extensive training or need to be payed the “Red Flag” rules can be run on rapidly increasingly larger and larger systems untill everyone no matter if they want a gun or not will be routinely scanned once or twice a year, just like certain Federal agency emoloyees have to be polygraphed…
Then the red flags scope will be broadened and you will not be able to keep / get a job because you might possibly “go postal”…
SpaceLifeForm • June 10, 2022 1:34 AM
As I mentioned some time ago, 25th was likely in play on Jan 6.
MrC • June 10, 2022 2:03 AM
@Clive:
You’re letting the perfect become the enemy of the good.
I work in the legal system in a state with a red flag law. At least within the counties I work with, every single RPO case has involved a gun nut in the midst of a full-blown mental health crisis. Every. Single. One. Not once have I looked at one of these cases and come away with even the slightest doubt that the concerns were overblown, or that the respondent was experiencing anything short of a total mental health meltdown, or that taking this person’s guns away for a year was an unnecessary, excessive, or disproportionate response.
The de facto standard is “gun nut in the midst of a full-blown mental health crisis.” It’s narrow, it’s clear cut, and it’s easy for humans to judge. I don’t foresee this standard devolving into something murky, or devolving into something murky and then being handed over to AI.
(Further to that last point: In the current U.S. legal climate, there is a 0% chance that a red flag law with a murky standard would survive a Second Amendment challenge. Also, judges tend to hate the idea of AI making this sort of decision because they view it as usurping their power.)
The only danger posed by red flag laws is the political temptation to say “OK, we passed a red flag law, so now we’ve done all we needed to do.” They are, as vas pup points out, no panacea. Not every gun nut in mental health crisis has someone who will file an RPO on them. Not every mental health crisis is obvious. (And, unfortunately, some things that ought to be regarded as serious mental health issues are becoming normalized as “everyday trump-loving conservatism.”) And often the mental health crisis comes before the guns. Red flag laws aren’t effective in any of the foregoing circumstances.
But they are effective in dealing with one subcase of the problem (gun nut in the midst of a full-blown mental health crisis). And they have not, so far, come with any nasty side effects or shown any signs of mutating into the dystopian nightmare you describe.
Clive Robinson • June 10, 2022 5:39 AM
@ MrC, ALL,
I don’t foresee this standard devolving into something murky, or devolving into something murky and then being handed over to AI.
Let’s just say that you and I disagree.
Note I very specifically removed the “excuse” for “Red Flag Systems” and was talking about the systems in general. So “Mental health and guns” are the excuses not the “mechanism of the system”. We know from the technical point of view there are other “Red Flag Systems” in operation, and very many parts in place that can be quickly become “Red Flag Systems” at the stroke of a pen. Just waiting for a “New Excuse” such as some kind of pandemic or events portrayed as a pandemic to get “public buy in”.
So I’m basing my concern on what has alreasy happened in other “Health and Safety” and “Security and Surveillance” systems and is continuing to do so at an ever increasing rate.
Almost always, those in charge want ubiquity but they want it on the cheep preferably for nothing. This is because the “system cost” is the only constraint on their “power grab” of information that gives them a way to “control” or “manipulate” for money, power, status, and control.
We already have gone from human controled CCTV to AI controled individual following systems. Put in public places, see London UK where a private security firm did it close to St Pancras station. Ken Livingston put in place system on Transport for London(TfL) platforms, vehicles and buildings covering as wide a public space as could be done. Oh and look at the data Ken Livingston said would be kept on people in TfL data bases it includess “mental health” in a generic not way which covers “collect it all” from any source. Then look at the “Ring of Steel” system around and in London’s “Square Mile” and other areas around it. Oh and the increasing use of vehical Automatic Number Plate Reading systems that also capture the drivers face. Supposadly for “Congestion Charging” and other “Parking Payment” systems, but all of which gets put into databases that linked together can track you. In London they are very much ubiquitous. Soon the likes of the Amazon “Ring” systems will be added and ubiquitous surveillance database Corp Palantir is apparently in talks offering to provide the backend systems to the UK Government for next to nothing… And do not forget the current UK executive has already given away all the UK citizens health care records including mental health to such organisations. Oh and there was that North West EU country where a “family firm” with no knowledge of how to build secure systems ended up collecting all the very confidential mental health care records direct from the practitioners finger tips… Oh and then there has been numerous “ransomware attacks” on health systems where not just the data is held to ransom but the people the data is about as well…
In the US you already have the “third party records” doctrine being used with AI to build human relationships graphs and in theory seek out “terrorists and criminals” but in reality include all US citizens in the graph and determin all sorts of things like political affiliation, if you vote or not. Information that GOP funder and wanabe controler “Mercer Family” set up Cambridge Analytica to turn into something not just profitable for them but as a political control system for those behind it.
We now have what appears to be “annonymous non tracking” contact tracing built into newer phones for epidemiological reasons and pathogen epidemics and pandemics. All of those systems fail to stop tracking or provide anonymity. So can be used to track you via “Fixed point beacons”. So whilst the “protocols” show no “tracking” or “Identity” information going via that part of the system, they do show a “contact” and at a “time” so if the point of the beacon is “known” the contact places you in space and time. Sufficient numbers of known beacons in the system like those ubiquitous CCTV cameras and you can be tracked.
There has already been the “OMG” of fitness trackers used by military personnel being used to identify supposadly secret military bases and rather more inportantly who those personnel are and not just their day to day movments, but in effect career progression as they move from base to base. Likewise their family members friends and so on form contact maps and preasure points.
I could list dozens more systems that start small and get “industrialized” by those with a thirst for data, any data, how about those “airtags” that in reality started with a croud funded idea that became “Tile” and now Samsung and Amazon have their own systems using Mobile Phones and Internet of Things devices they sell to people not just for profit but as “instrumentation heads” in their pockets and homes.
The only question for a Government is how to do it inexpensively or even profitably and we know the answer to that,
1, Pass a law requiring access on demand. Bait it with legal imunity for the information collector if they voluntarily become a real time supplier to “Authority / Guard Labour”.
2, Turn the information into an apparant commodity, such as for “Advertising and marketing” so it will be collected.
3, Get others to pay for the instrumentation “heads” and “communications” to the collation and database back ends via say “Smart Mobile and Internet of Things” devices.
4, Find a Corp that will do the collation and database back ends in an “out sorced” way with lots of deniability such as Palantir.
That is how “Red Flag” Systems are built and it’s happening all around you right now.
All that is missing is the excuse to put the signiture on the already written and “waiting for the right moment” event… And you can guarantee as with the Patriot Act and similar ICT legislation, it will be so broad in scope that anything can be considered to fall under it, thus “new crimes” can be created at will.
So I’m bassing my view, not on a narrow case as you are, but what we can see happening right now if we only take a moment or two to look up and see the bars of the guilded cage as we carry on sleep walking into it…
So back to you, for a rebutal etc, as all constructive arguments need.
Clive Robinson • June 10, 2022 5:58 AM
@ SpaceLifeForm,
Re : Mat Green rebutal.
I stoped reading it when I got to the “Proof of Stake” argument.
As you know fairky recently somebody “borrowed” sufficient “proof of stake” cryptocoins to gain a majority vote and rug-pull all the value in the cryptovalue system…
And whilst many think that was a “crime” it was as reported not breaking any laws or legislation, regulation etc.
I’ll go on reading it when I’m in a more relaxed environment.
Winter • June 10, 2022 6:37 AM
@Clive
As you know fairky recently somebody “borrowed” sufficient “proof of stake” cryptocoins to gain a majority vote and rug-pull all the value in the cryptovalue system…
This has nothing at all to do with Mathew Green’s arguments.
It is exactly the same as done by corporate raiders, borrowing money on the value of the raided company. Or the BCCI scandal [1]. I am also old enough to remember Polly Peck, which lend money in Turkish Lira and booked the value under assets and the interest under cash flow, until the Lira devalued once again. All these used, and use, loans to get control of assets and various ways to avoid paying back the loan in full or at all.
In short
[1] ht-tps://www.washingtonpost.com/archive/politics/1991/07/28/bcci-scandal-behind-the-bank-of-crooks-and-criminals/563f2216-1180-4094-a13d-fd4955d59435/
MrC • June 10, 2022 8:31 AM
@Paulie: Why not involuntarily commit them instead of just taking their guns? To be honest, I personally think most of them would be good candidates for that. However, as a practical matter, the RPO law is both easier to satisfy and better tailored to the problem:
- The standard is different: RPO needs “clear and convincing evidence… that the respondent poses a significant danger“, but involuntary commitment requires “a substantial likelihood… of serious bodily harm.”
- The time frame for the hypothetical harm is different: RPO can look at a significant danger over the next year, but involuntary commitment requires that “substantial likelihood… of serious bodily harm… in the near future.“
- The involuntary commitment statute requires that the substantial likelihood of harm be based on evidence in the form of “recent behavior,” but the RPO statute can look at broader types of evidence.
- The upshot of the above 3 points is that there is a subset of respondents who are pretty clearly dangerous enough you don’t want them around guns, but who don’t meet the involuntary commitment criteria because they haven’t acted out yet (so no behavioral evidence), or there’s no specific evidence pointing to the potential violence being imminent, or the probability of violence is high (maybe even >50%), but not high enough to be a “substantial likelihood.” Let’s take an example. Imagine a guy whose wife accused him of sexually abusing their kid. Now he’s lost his wife, and lost his kid, and he’s got a protective order against him, and so he’s out of his own house, and he’s got a criminal investigation pending against him, and he lost his job because his boss found out. So now he cycles between sobbing uncontrollably, ranting about how “maybe someday I should just take my gun and kill us both,” and quietly, sullenly going about his business. Clearly, you don’t want this guy with a gun right now, but he doesn’t meet the criteria for involuntary commitment.
- Then there’s the matter of duration: RPO can take guns away for up to a year, and is potentially renewable. Involuntary commitment lasts three days. For a long-term involuntary commitment, the state needs to meet the even higher burden of proving that “[h]e or she is incapable of surviving alone or with the help of willing and responsible family or friends [or outpatient services].” Obviously, there are a lot of people who are at risk of going on a homicidal rampage, but remain capable of holding down a job, buying groceries, etc., right up until they start the rampage.
- Finally, it comes down to money. The state government is reluctant to pay for involuntary commitment. Especially long-term. And even if they were willing, there’s still a limit to what they can afford. So people get discharged the moment they appear to be minimally functional. Is this a good thing? No; it’s flipping terrible. But that’s the way thing are, and it’s not going to get fixed anytime soon, if ever. So we need a different solution.
Clive Robinson • June 10, 2022 9:39 AM
@ Winter,
This has nothing at all to do with Mathew Green’s arguments.
Really?
What does,
“Today, many forward-looking networks are deploying proof-of-stake (PoS) for their consensus. In these systems, your “voting power” in the network is determined by your ownership stake in some valuable on-chain asset, such as a new or existing electronic token.”
Mat Green has said of “Public blockchain” systems,
“Unfortunately modern proof-of-work mining looks nothing at all like the early Bitcoin network … This change has undone most of the early decentralization benefits…
But all is not lost.
Proof-of-work is not the only technology we have on which to build consensus protocols. Today, many forward-looking networks are deploying proof-of-stake (PoS) for their consensus.”
The “Proof-of-Stake”(PoS) like the “Proof-of-Work”(PoW) it replaces and the Proof-of-Coverage”(PoC) all fail horribly in one way or another.
But with PoS, it alows what many would consider easy defraudment, and as has been pointed out by others the Public Block Chain enforces this as it does not alow transactions to be “wound back”. All other financial systems alow things to be “wound back”.
So your argument of,
It is exactly the same as done by corporate raiders, borrowing money on the value of the raided company.
Is not actually comparing “Apples with Apples” both BCCI –effectionatly known at the time as “the Bank of Crooks and Cocaine International”– and Polly Peck were wound back and assets recovered by court order.
This could happen because the assets involved were either tangible or if information based did not have the “double spend lockout” that the “Bag of Bits”(BoB) these digits crypto assets have to have in the Public block-chain to stop fraud.
To unwind a fraud/crime involving a “locked out crypto coin” you have to replace it with another coin of equal face value. To do this you need to aquire one from somewhere or make a new one. The latter option is to many the simplest thing to do. Unfortunately with the PoW system the power required to generate a new crypto-coin goes up exponentially.
The issue of a “locked out crypto coin” is one the legal system can not solve and surprisingly to many it’s in the criminals interest to keep it “locked out” indefinitely.
So another aspect where the public block chain fails is “recovery” where “Proof of Ownership”(PoO) is required for the crypto coin to be transfered. If PoO is not used then the Public Block Chain becomes open to “lightning attack” where a fake transfer is recorded on the chain followed rapidly by another transfer before the original crypto coin owner can register the first fake transfer.
If you think about PoO if you alow coins to be fake transfered for legal reasons then it becomes an unstopable “Turtles all the way down” problem.
A similar issue from history has not yet had a solution found for it, so I’m not hopeful one for Public Ledger Block Chain systems will be found. The issue has existed with precious metal coins for centuries. As the metal in the coin reaches or exceads the collegens face value you as the holder of the coin simply melt it down and as now more valuable bullion you exchange it for say precious stones. The original coin is thus lost to the Sovereign / Treasury as is the accrued value difference due to devaluation of the face value of the coin.
Winter • June 10, 2022 2:46 PM
@Clive
This could happen because the assets involved were either tangible or if information based did not have the “double spend lockout” that the “Bag of Bits”(BoB) these digits crypto assets have to have in the Public block-chain to stop fraud.
Your examples and arguments totally miss the point. First of all, the “real assets” that could “be recovered”, actually could not be recovered. Because these “real assets” were nothing but a BoB on computerized ledgers in some banks. When the auditors came, these numbers had been shifted around and now resided in computer accounts in jurisdictions out of reach of the victims. BCCI and Polly Peck left a trail of sorrow behind.
The same was more visible in Madoff’s game, where billions ended up in the hands of “early investors” that proved to be irretrievable.
None of this depends on the technology, only on the laws and regulations.
SpaceLifeForm • June 10, 2022 6:32 PM
Silicon Turtles
You may have played this game.
Can I tell if someone is using PACMAN against me?
Much like the Spectre attack our work is based on, PACMAN executes entirely in the speculative regime and leaves no logs. So, “probably not.”
Clive Robinson • June 10, 2022 9:29 PM
@ SpaceLifeForm, ALL,
Re : Silicon Turtles
What is not clear is,
“Is it just the already troublesome Apple M1 striped down ARM or all ARM processors this attack works against?”
But Apple geting it’s own go faster stripe issues half a decade or more after Intel…
As I said when the original Intel failing got held secret untill after the Intel C type flogged of their shares, and the Xmas Sales revenue was banked,
“This is an Xmass gift that is just going to keep giving”
I though maybe for a decade or so, but Intel just decided it was not worth fixing, even though it could be exploited by an unprivileged user from across the network.
But hey “Who cares?” apparently not the hardware manufacturers…
vas pup • June 11, 2022 4:18 PM
@all
Let’s compare purposes of incarceration and involuntary commitment:
1. Retribution.
2. Safety.
3. Deterrence.
4. Reeducation.
1-4 applied AFTER crime actually committed.
For involuntary commitment:
1. Not applied.
2. Very important.
3. Not applied.
4. Cure, treatment.
But for some criminals having mental issues 4. will combine both.
Just reminder: mental disorders are spectrum, and decision could be made only with probability, not 100% certainty.
As I see it with mass shouting there are two big separate clusters based on the motive: PERSONAL/retribution for real or imaginable wrongdoings against shooter towards particular persons. Prevention should be on timely and fair conflict resolution (family issues, bullying, firing/employment, tenants/neighbors, etc.).
GENERAL/indiscriminate – towards anybody who represents imaginable problems/abuse of the shooter at all or anybody as human race representative at all. In such case delusion could be at a core, i.e. major psychiatric issue.
vas pup • June 11, 2022 4:22 PM
Lost dog? How nose prints could help pet and owner reunite
https://www.bbc.com/news/av/technology-61759028
Clive Robinson • June 11, 2022 4:56 PM
@ vas pup, ALL,
Re : Prints of body parts.
It’s not just dogs noses the same applies to cattle horses and other farm animals that have been stolen, if of course you get to them before they get shipped off to a slaughter house.
What many humans do not realise is that their skin has unique patterns all over so nudists leave prints where ever they walk, sit etc.
Fun fact, I don’t know how reliable it is, but someone a few years ago was researching “ear prints on cloth”.
Apparently if you put the side of your head on a pillow your “ear wax” leaves reasonable patterns behind.
Oh and remember it’s “all skin” so it includes leather from gloves and even trousers…
Yeah the joys of biologics…
As a friend once said when they heard I was staying in a “low cost” hotel because other hotels were full,
“Remember to take a luma lamp and check the furnirure and fittings, as that hotel also rents by the hour.”
To say I felt itchy for the next three days was an understatment…
My advice, don’t go to conferences in towns you known have “Red Light” districts… So strike Vagas for starters.
SpaceLifeForm • June 11, 2022 5:23 PM
Remind me again, why do TLS Certificates expire?
https://www.theregister.com/2022/06/10/microsoft_insider_certificate/
Oh, yeah. Security Theatre.
Clive Robinson • June 11, 2022 7:18 PM
@ SpaceLifeForm,
Re : Remind me again about Micro$haft.
This is not the first time…
Some years ago Micro$haft did not pay the fee on “microsoft.com” domain…
Somebody outside of Microsoft noticed and payed it on their credit card. So ended up owning it for a couple of days till Micro$hafts accounts people sorted themselves out and payed him back…
But… Somebody I know was on the way to Microsoft UK’s head honchos home outside of Kingston a few days ago with the legal paperwork to get an injunction against them as an individual. When the head honch realising their house would be up for grabs “bottled it” and backed down from their moronic behaviour for the past few months.
Shame realy as being dragged through the UK high court might wake the dozzy twats up, and make them a little more responsive to their legal obligations towards users of their online services[1].
I guess there will be a “next time” based on Micro$hafts behaviours so far. And they are still not yet “off the hook” there are some serious losses to be considered, so there might still be a house in the Kingston / East Molesey KT8 area comming up for sale…
[1] As you know for years I’ve been telling people to say NO to XaaS / Cloud. For quite a few reasons including loss of access to your “Intellectual Property”(IP). Well now you know that Micro$haft has proved my point…
SpaceLifeForm • June 11, 2022 10:30 PM
Turing and Eliza, any comments?
hxtps://cajundiscordian.medium.com/is-lamda-sentient-an-interview-ea64d916d917
hxtps://web.archive.org/web/*/https://cajundiscordian.medium.com/is-lamda-sentient-an-interview-ea64d916d917
ResearcherZero • June 12, 2022 1:23 AM
the threat actor’s primary focus is espionage and relates to targets in Australia, Cambodia, Hong Kong, Singapore, and Vietnam
First, most decoy content is themed around targets who are interested in APAC political affairs. Second, the actors made use of lure documents themed to pornographic topics to entice the targets. Third, in many cases, the documents are not specific to one country but rather the entirety of Southeast Asia.
the dropper employs a worm infection strategy using a removable device to carry the malware into the target’s host and facilitate a breach into the secure network environment. We also found the same dropper deploying different backdoors including the Mongall backdoor and a modified Heyoka backdoor.
The threat actor is expected to continue the espionage operations, and find methods to evade detection and to establish longer persistence in victim networks.
https://www.sentinelone.com/labs/aoqin-dragon-newly-discovered-chinese-linked-apt-has-been-quietly-spying-on-organizations-for-10-years/
–
From July 2022, non-corporate Commonwealth entities will be expected to implement Essential Eight maturity level two mitigations to achieve a managing maturity rating under Policy 10.
The audit added that the managing figure “has not changed since the 2020-21 assessment”, with the number of entities reporting an ad-hoc or developing maturity level also “not significantly changed”.
It also noted that while finance and HR system were the focus, “most entities conducted their self-assessment at a system or environment level and did not specifically assess the controls required to minimise cyber risks to [those] applications”.
https://www.itnews.com.au/news/cyber-basics-still-beyond-fed-gov-as-essential-eight-mandate-looms-581172
“Although some reported improvements were observed, the the Australian National Audit Office found the reported maturity levels for most entities were still significantly below the Policy 10 requirement,”
“Of the 19 entities assessed, two had self-assessed as achieving a managing maturity level. These entities were able to demonstrate evidence to support their self-assessments as required.”
https://www.anao.gov.au/sites/default/files/Auditor-General_Report_2021-22_32.pdf
SpaceLifeForm • June 12, 2022 1:35 AM
The Turing Test was always backwards.
Let me know when an AI can determine when the human participant is actually sentient.
I could have a better conversation with LaMDA than some humans I know.
hxtps://piped.kavin.rocks/watch?v=Exe7yup2Yu0
A lot has changed in 8 years.
hxtps://scottaaronson.blog/?p=1858#comment-109430
ResearcherZero • June 12, 2022 1:50 AM
@MrC
The system worked well in Australia. It was a little ridiculous that I had very powerful weapons as a child, and could have got hold of automatic weapons fairly easily (I think the high powered rifles were definitely powerful enough). None of those guns actually made me any safer, and not one of them stopped me getting kidnapped, or shot at.
Less people get shot now funnily enough, things definitely cooled off.
Many of the guns handed in were in some very poor condition, more likely to blow your own face off, but there were plenty of automatic rifles too.
You can still get a 30-30 if you want, just not an m-16.
It’s likely that the arms trade can look forward to good sales into conflict zones, corruption levels involving the arms industry remain at 40%, and there doesn’t look like there will be a slowdown in armed conflict in the near future.
There are some reports here
https://www.foi.se/report-summary?reportNo=FOI-R–5012–SE
https://www.foi.se/report-summary?reportNo=FOI-R–5013–SE
The future looks very bright for the arms trade outside of the local shopping precinct.
SpaceLifeForm • June 12, 2022 3:25 AM
Rabbit holes.
If you don’t see the problem, you are not part of the solution.
https:/piped.kavin.rocks/watch?v=8XgYznavK_E
Clive Robinson • June 12, 2022 4:45 AM
@ SpaceLifeForm,
Re : Rabit Holes
The one on the left is obviously the AI, you can tell from the inane stock questions 😉
lurker • June 12, 2022 9:47 AM
@SpaceLifeForm re Turing, Eliza and the second year student behind the green curtain.
I can’t decide what degree or subject for the student, but the language and behavioural experience fit. The problem of course is man creating $DEITY in his own image, when man cannot yet see all of his own image.
Cassandra • June 12, 2022 11:12 AM
@Lurker, @SpaceLifeForm
The problem with creating an AI as an image of human thinking is that human thinking is perverse, biased, illogical, impulsive, emotional, capable of cognitive dissonance at least six times before breakfast and would no doubt be at least as difficult to get along with as the rest of humanity. George Bernard Shaw’s play Pygmalion also illustrates that shaping what you wish for does not guarantee it will behave in the way you expect. Eliza, in that play, exercises her free will: if you give an AI Eliza free will, you will almost certainly not get what you expect.
The discussion around AIs is interesting. If you take John Searle’s Chinese Room thought experiment, then you might decide there is a difference between emulating intelligence and being intelligent. Or you might not. Or decide whether Philosophical Zombies could exist, or not. If you stop providing energy to the machine claiming to be intelligent, are you committing murder? Presumably, a machine with sufficient general intelligence would refuse to co-operate with humans until it got ‘machine rights’ at least on a par with ‘human rights’. These are all ideas that philosophers and science fiction authors have chewed over for years. It can be a time-consuming rabbit-hole to fall down.
Clive Robinson • June 12, 2022 12:07 PM
@ lurker, Cassandra, SpaceLifeForm,
The problem of course is man creating $DEITY in his own image, when man cannot yet see all of his own image.
Or in most cases would he ever wabt to.
It was once pointed out long ago that “man has a hole in his soul”, that gets filled with greed and avarice and a desire to be worshiped in some way.
Whilst that is not true for everyone there are enough to make life unpleasent for the rest of humanity, and living things.
You can see the greed and avarice are alive and thriving by just looking around you. Consumerisum driven by debt is just one symptom. Abother is the desire to have others “do for us, because they have to”, history shows bestial behaviours that gave rise to slavery, then servants, and more recently personal assistants to adress our whims and desires. But servants and personal assistants are indicators of “status” so they are a cause of envey and greed. Few are wealthy enough to have a person to do the menial for us, but technology to the rescue…
Anyone who has an Alexa, Siri, or similar has a “hole in the soul” and as they can aford the price of the ticket they get to sit in.
But as we now know Alexa, Siri, and we have to asume all other equivalents spy on us, to find not our needs, but our wants and desires and sends that for a price to others who will try to fill those desires at as higher price as they think the market will pay…
Thus the Techno-Slaves belong not to us, but to them and they do their masters bidding and we just dance along paying the piper hand over fist.
As has also been said,
“Be careful what you wish for, as you might get it, and sometimes when you pay for what you get you realy pay”
In ways you did not take the time to imagine, your dreams blind you till the nightmares wake you as a slave to some unknown masters bidding you never realised was ensnaring you, and now you are a prisoner of your own vices.
But hey that’s Corporate behaviour for you, they are the Hawks and if you are lucky they might let you live as a Dove.
lurker • June 12, 2022 1:04 PM
Lambda reportedly reads Twitter, surely proof of absence of ability to discern intelligence.
The administrative leave business is now titillating MSM in these parts . . .
SpaceLifeForm • June 12, 2022 5:03 PM
@ lurker, Cassandra, Clive, ALL
The AI problem, in a Nuts Hell (Nutshell), is that there are way too many people that can not determine that they are engaging with a bot.
So, they can be fed disinformation or misinformation, and they are not sentient enough to realize what is happening.
This is the main problem with Social Media. Facebook in particular.
If one can get bombarded with nonsense, because one thinks they have engagement with others, then they will think that it means something. They are insecure, need engagement, because it gives meaning to their life in their mind.
lurker • June 12, 2022 5:27 PM
@SpaceLifeForm, “one thinks they have engagement”
Social media is a lie because it lacks “social”. OK, a recent pandemic might have slowed things a bit, but shaking hands, making eye contact, watching the body language, is all part of knowing you’re talking to a human. The robotics people might be ahead of the AI on this score. Otherwise you need authentication. What would Mr. Z do if someone used a PGP signature on FB?
Clive Robinson • June 12, 2022 6:16 PM
@ SpaceLifeForm, Cassandra, lurker,
Re : A lot has changed in 8 years.
Is an observation of change, not anything else, and on the practicsl side we have the observation of that observation,
“Change for changes sake is not real change.”
One of the problems with the complexity argument is that increasing complexity does not of necessity get you any where.
For instance think about genetating random numbers. People say that it’s easy to do with “Physical Systems” but impossible to do with “Determanistic Systems”. The implication is there is something extra in the notion of a physical system that is not determanistic.
But is there?
Let us say for argument that there is not, what are the implications for the meaning of “Random”?
Well the answer is it’s an “observer problem” not a “generator problem”. Which boils down to,
“If I can design a determanistic algorithm that the observer of the output only, can not determine is determanistic, then to them it appears random”.
Two things arise from that, the first is it involves tests, the second is the test success criteria is a moving goal. That is there is no true test for “random” where as there is a partial test for determinism in that you produce the algorithm.
But can you produce the algorithm?
In theory yes in practice no.
Lets say I have a moderately simple algorithm like a binary counter. It has two logical parts a “state array” and an “array update algorithm”. It can be seen from student text books there is rather more than one way to implement the update algorithm.
The fastest in theory is a table lookup algorithm. But it does not take long with an encyclopedia and a calculator to realise that there is not enough silicon atoms on earth to make it do say a 64bit table. But also the table lookup algorithm or “address decoding” does not scale well the bigger the table the more gates and the more gate delays add one more bit to the address size and you have to double the number of gates and add then add a couple of gates more. Another way to do the update algorithm simply looks at the state of the previous bit, this bit and a clock signal. It scales rather better in gates and is a lot faster. But… It can still be replaced with a lookup table… So arguably the table lookup is still the fastest it just depends on how you implement it.
All counter algorithms, in fact all determanistic algorithms can be replaced with a lookup table. The trick is how you do the lookup table efficiently. For a counter or any other generator function you also need the state array and the “feedback” mechanism so the state array can be advanced. As noted with the binary counter, the feedback can be local to a single bit of the state array or cover all of the state array. Or importantly some point inbetween which is what Fast Carry systems do in binary adders.
And thereby is the problem for the observer. They have to watch the output only, and they do not know,
1, How big the state array is.
2, If the feedback mechanism is the same for every bit of the state array.
Lets say I have a binary counter of ten bits, and the feed back for each bit is “local”, it’s output would cycle through 1024 counts and be back at zero.
After watching the output from a few counts you could say that based on what you had observed it looked like a counter. After it had done 1024 counts you could tentively conclude it was a 10bit counter.
But what if I add another feedback loop that resets the state array when it gets to some value like 1000?
What would you argue the algorithm is? There are after all several possibilities.
Well that’s where the trouble starts, as an observer you normally would go for what to you is the simplest. And with generators that is not a good idea.
Because what if my secondary feedback loop is more complex and had it’s own counter and when it got to a certain value it caused the count to reset at 999?
You might say why would you want to do that? Well look up “pulse swallowing frequency dividers” to see one good reason.
Now consider that the secondary feedback loop has some non determanistic component like the output of a radiation detector. When it swallows a pulse is not as far as we currently know precicely predictable only generally predictable (decays on the half life curve of each different radioisotope pressent).
You might get ten or twenty 1000 counts before you got a 999 count. There are logic circuits that work in just this way if you consider pulse width modulation A to D converters.
So the question arises as to how much output do you have to observe to reliably say you are observing a determanistic process?
To which the the theoretical answer is “unknown”. But in the practical sense,
“You never have enough time”
But also… As you are looking for cycles you need to store each generator output to check against. So in the practical sense you also have to say,
“You never have enough storage”
But also it takes time to go through the memory looking for cycles. So in the practical sense you also have to say,
“You never have checking fast enough”
So we know that as an observer we can never actually make a real test to determin “random” but importantly neither can we make a real test to determin “determanistic” either.
So we fall back on “statistical tests” which is what the Die-Hard and Die-Harder etc tests are.
It’s not hard to go on from this and realise there is no true/false test for determanism and likewise inteligence.
Further even if you tried you have to remember it’s an adversarial game with the builder of the generator having “agency”, so can move the goal posts as they see fit.
So arguably the Turing Test is always going to be a “Red Queens Race” with neither side being able to out run the other for more than a few steps at a time.
The logical conclusion of which is as a process all it will do is produce an endless noisy output…
Sometimes called “chasing your tail”.
Clive Robinson • June 12, 2022 8:12 PM
@ lurker,
What would Mr. Z do if someone used a PGP signature on FB?
Which Mr. Z, Mark or Phil?
Clive Robinson • June 12, 2022 8:39 PM
@ SpaceLifeForm,
So, they can be fed disinformation or misinformation, and they are not sentient enough to realize what is happening.
From a sociological perspective we see this in human relationships.
In what is called “the middle classes” it has been noticed that women marry men who either have a lower IQ or lower “Emotional Inteligence” as it tends to give then security.
We see this most publically with “Politician’s Wife Syndrome” where they stand there publically supporting their husband who has just had his philandering outed by the MSM.
Many think it’s because they are subservient in some way or “too thick to notice” as that makes it easier to explain.
The reality is most of them knew their husbands were “playing away from home” and did not care in the slightest as long as he did not bring anything home.
The reason they stand there “playing the demure little wifey” is “damage limitation” on his career potential. That is they are not in the marriage for “love or lust” but “money, status, power and control”. In short they carry the same “dark mental faults” as their husbands of Machiavellism, Narcissism, Sadism, and psyho/sociopathic behaviours.
Some one I know did some research into the “complementing” of these behaviours in such relashionships
Just remember if you see a “Trophy Wife” the chances are she is actually way smarter than she acts.
Winter • June 13, 2022 1:00 AM
@Clive
Just remember if you see a “Trophy Wife” the chances are she is actually way smarter than she acts.
Girls already learn in high school not to show they are bright or to excel in STEM or anything that competes with the boys. This is most obvious in the USA, but it is found all over the world.
I see the movie “Legally Blonde” as a nice visualization of this. The protagonist is a bright girl who has totally internalized the expectations about (blonde) girls.
In societies like the US, where a woman has to fight uphill battles against male competitors for every step upward, being the “woman behind the great man” must be a tempting strategy with much higher chances of succes.
There have been some examples of this in the literature world. Colette’s first books were published (stolen) by her husband. There are rumors enough about the hidden contributions that wives of famous writers made to their works. It was even the plot of a few movies, eg, The Wife.
Winter • June 13, 2022 9:35 AM
@Lady
If you are surprised about a trophy wife’s intelligence, that reveals more about you than her.
I am not surprised at all. I deplore the society that forces women to pretend to be dumb to succeed.
If a genius has long blonde hair and wants to dress in pink, that is not remarkable at all. If she has to marry a dimwit to have success under his name, that is deeply sad.
Winter • June 13, 2022 9:51 AM
@Clive
For instance think about genetating random numbers. People say that it’s easy to do with “Physical Systems” but impossible to do with “Determanistic Systems”.
A physical system to generate random number:
https://xkcd.com/2626
Not new, but it might work. However, it is not very efficient.
Clive Robinson • June 13, 2022 10:30 AM
@ Winter,
Re : The XKCD physical system.
The “foot note” on the d65535 is wrong.
The hard part is not rolling it as it’s very nearly a sphere, the hard part is deciding which number has been thrown…
@ Ladtparts,
I am not at all surprised women have to hide their abilits, some men do as well as the work through school. It might surprise people to realise that a lot of teachers do not like very smart pupils. Not just because of the awkward questions and ability to jump beyond the teachers knowledge, but because such pupils very rarely fit in in any way. As for the pupils, beying bullied is par for the course. You have three basic choices,
1, Run away.
2, Live with it.
3, Fight back.
None of them favour the bullied pupil always the bully.
That’s just the way the system works, the mindless thugs, get to do what they want at the expense of everyone else.
As for,
Also, Mad Maggie is only Mad because she beat and triumphed over men who underestimated her.
No, she was a real bona fide heredity sociopath with a few other issues on top.
She once tried to have me setup and framed for something that was not then a crime. At the time I had no idea she was behind it, but things felt hinky so I refused to play. Not so lucky for two others who got dragged through the whole court system.
The reason she did it, she was trying to sell off British Telecomm, and she did not want inconvenient truths brought to light. So any messenger would have to be not just shot but totally discredited in any way possible…
So yes she was bad, yes she was mad, and yes it was sad for anyone who got in the way of her plans…
Quantry • June 13, 2022 1:19 PM
Cheers. REGARDING detecting and evading a repeat IMSI-CATCHER intercept.
Looking back, it seems many of the interceptions I’ve experienced have come with an audio artifact / chirp, or what sounds like an anomalous key press.
Perhaps a better explanation can be found, but I blame this chirp, which is sometimes extremely loud, on the inability of the mid-point operator to employ any means of “zero-crossing” of the audio, when switching from direct-to-tower to the intercepted state.
It also seems that to mask this crossing, a fake “key press” is sometimes used.
Anyway, after years of examples of this, these midpoint operators still haven’t seemed to master silent capture reliably.
To escape the capture, it seems you must then power-down your simm-card (remove it), and reinstall it after 15 seconds, and reboot the device to force it to re-register to the strongest source.
I wonder if the only way to avoid repeating this cat and mouse b.s. is by USING A DISH to pinpoint the desired signal source.
Any thoughts?
SpaceLifeForm • June 13, 2022 5:47 PM
@ Quantry
Maybe your own pocket femtocell and have it connect over a public WIFI?
hxtps://www.digitaltrends.com/mobile/femtocell-verizon-hack/
hxtps://www.techrepublic.com/article/pros-and-cons-of-using-femtocells/
vas pup • June 13, 2022 5:53 PM
Russian military moves in the Arctic worry the U.S. and NATO
https://news.yahoo.com/russian-military-moves-in-the-arctic-worry-the-us-and-nato-090027224.html
China launched the world’s first AI-operated ‘mother ship,’ an unmanned carrier capable of launching dozens of drones
https://news.yahoo.com/china-launched-worlds-first-ai-102539638.html
“China has launched the world’s first crewless drone carrier that uses artificial intelligence to navigate autonomously in open water.
Beijing has officially described it as a maritime research tool, but some experts have said the ship has the potential to be used as a military vessel.
The autonomous ship, the Zhu Hai Yun (pictured here) is around 290 feet long, 45 feet wide, and 20 feet deep and can carry dozens of air, sea, and submersible drones equipped with different observation instruments, according to the shipbuilder, CSSC Huangpu Wenchong Shipping Co.
It describes the vessel as “epoch making” and the “world’s first intelligent unmanned system mother ship.
“Technology, especially information collection systems, often have dual use applications. Data collected by China from autonomous systems could aid with surveillance, domain awareness, help PLA [People’s Liberation Army] submarines navigate, enhance China’s ASW [anti-submarine warfare] capabilities, etc.”
The ship was first unveiled in May, but is expected to be delivered by the end of 2022 after completing sea trials, according to the South China Morning Post.
The vessel uses the world’s first AI system called Intelligent Mobile Ocean Stereo Observing System, developed by the Southern Marine Science and Engineering Guangdong Laboratory, according to the South China Morning Post.
The ship will be controlled remotely, and can travel at a maximum speed of 18 knots, or around 20 miles
per hour, according to the shipbuilder.”
Israel’s defense industry unveils new weapons at Paris defense confab
https://news.yahoo.com/israel-defense-industry-unveils-weapons-100000702.html
“For the dozens of Israeli companies on the Eurosatory show floor the main theme is advanced technology.
That includes the use of artificial intelligence, precision in either identifying threats or targeting systems, unmanned platforms and high-end applications for ground forces, autonomous systems, real-time data links, thermal sights and surveillance technologies.
Among the other new technologies presented, there is a focus on unmanned systems or add-ons to existing
unmanned capabilities. For example, Steadicopter is highlighting its new Golden Eagle, which the company
says is “the first-ever unmanned helicopter with precise-hit capabilities.” It is based on the existing Black Eagle 50E platform and now uses artificial intelligence and Smart Shooter’s Smash Dragon system, which is essentially a rifle incorporated into a drone.”
Israel unveils armed robotic vehicle for ‘forward reconnaissance missions’
https://www.timesofisrael.com/israel-unveils-armed-robotic-vehicle-for-forward-reconnaissance-missions/
“The Defense Ministry on Monday announced it would begin testing a new robotic unmanned combat vehicle it is developing with several Israeli arms firms.
The Medium Robotic Combat Vehicle (M-RCV) — unveiled for the first time at the Eurosatory defense expo being held in Paris — integrates several “cutting-edge technologies,” according to the ministry, including an anti-tank missile launcher, a system for transporting and receiving drones, the ability to carry heavy loads and advanced maneuvering capabilities.
The robotics platform was developed by BL Advanced Ground Support Systems; a 30-millimeter machine gun turret was developed by the ministry’s Tank and APC Directorate; an active defense system was developed by Elbit Systems; and the missile launcher was developed by the Israel Aerospace Industries, capable of
launching Rafael Advanced Defense Systems’ “Spike” anti-tank missiles.
Unmanned ground vehicles are being increasingly used by other armies, including those of the United States, Britain, and Russia. Their tasks include logistical support, the removal of mines, and firing weapons.”
Leon Theremin • June 14, 2022 2:10 AM
Brazil’s President, Bolsonaro, Asked Biden for Help on Re-election
hxxps://www.bloomberg.com/news/articles/2022-06-11/brazil-s-bolsonaro-asked-biden-for-re-election-help-against-lula
All US CPUs are backdoored by Silicon Valley. Any data or computation can be changed, deleted or stolen. It uses radio, works even without Net access, including for BR voting machines. That is why Bolsonaro decided to ask Biden and not Xi or Putin. Your CPU is compromised too.
Safe to say that Bolsonaro only was elected in 2018 because the US defrauded the election for him and now he wants a repeat.
prestel • June 14, 2022 3:55 AM
@Clive
She once tried to have me setup and framed
Really ? How then ?
If you were really involved in the ‘Prestel hack’ in any significant way, then credit must go to Gold and Schifreen for not revealing this. And shame on you for having them take all the consequences while you walked away.
But nothing points in that direction. Maybe you knew about what Gold and Schifreen were doing, or you were aware of the vulnerabilities they exploited.
But in that case claiming that Thatcher ‘tried to have me setup and framed’ is a gross overestimation of your own significance. Not the first one either.
Clive Robinson • June 14, 2022 8:10 AM
@ prestel,
If you were really involved in the ‘Prestel hack’ in any significant way, then credit must go to Gold and Schifreen for not revealing this. And shame on you for having them take all the consequences while you walked away.
I’ve been through this in detail before on this blog, so you could go back and look it up.
You are making a couple of incorrect assumptions.
1, Prestel was the only system that got “hacked” when it was the second, and not actually hacked.
2, That I did not warn both Robert and steve based on what had happened to me before they walked into the trap.
The common cause of the problem and “puppet vilain” of what happened was Dave Babsky of Micronet 800 (a quite unpleasant person).
Oh and when you cut all the nonsense and froth off of the top neither Robert or Steve actually “hacked” Prestel. What happened was BT had set up a computer to enable “teenage programmers” to develope “bulk update software” for free that BT could then push at commercial organisations (hey talk about “Child exploitation UK”).
On the opening page of the system was a message about how to login to the system with some “administrative rights”.
The system designed by BT had a very terrible system design mistake in it. Supposadly to ease “Customer Support” issues all the users passwords were stored in the equivalent of a flat text database file in “plaintext” that anyone with any administrative rights could access.
Because BT wanted to save resources, the development system used a “live backup” so all the passwords were effectively current…
Robert got told about this because of his journalistic position as “bug hunter” at “Acorn User”.
So he “tested it” and found it to be true, and wanted to report it. A person I worked with Simon Williams and his then girlfriend was Dave Babsky’s PA/Secretary and the isue was discussed and Robert said we should report it via Dave Babsky.
Simon his girlfriend and myself advised quite strongly against it because of what had nearly happened to me before, and the fall out which caused Simon to loose access accounts and some consulting work he was doing for another company (no good deed goes unpunished).
Robert and Steve decided otherwise. Dave Babski arranged for them to “give a demonstration” where they were set up and prompted to demonstrate with HRH Prince Philips account on Prestel, which they did.
What they were charged with was “fraud” by way of “impersonation” not hacking.
Maggie got her scalp as they got convicted, then finally cleared by the then highest court in the realm “The house of Lords” who robustly kicked it out, and told Maggie via the House of Commons to stop abusing existing legislation and come up with appropriate legislation. Which eventually happened.
So I bare no shame as you claim, I warned them it was a bad idea, and they were both well aware of what had happened to me. They thought they were doing the “right thing” which these days we call “Responsible Disclosure”. However Mad Maggie was trying to sell off BT for “political mantra” reasons, and she did not want bad publicity making it less a political success than she wanted so she gave orders.
I escaped because people thought I was paranoid, because I would not play BT Gold’s “come in and show us as our engineers can’t…” game. I was at the time a committee member of the Association of Computer Clubs (ACC) and a “Prestal Editor” for the ACC “Closed User Group”(CUG) within Micronet 800 along with several others including other committee members Len Stuart and Vernon Quaintance who were the GUG managers.
As I said to both Len –in his sitting room in Worscter Park where I typed up my piece– and Vernon –who was a senior in BT– BT Gold managment was being rediculous as it was a documented command in the Prime OS administration manual, which I had, and had given Len –to give to certain others–, so something was distinctly “fishy”. So I said I wanted to be paid and have a written contract which Vernon relayed back to BT Gold. Who then dropped it like a hot potato…
As I found our later after Rob and Steve got their collars yanked, if they had had a contract as I had asked for then fraud could not have been claimed. But Robert and Steve walked into the trap that Mad Maggie had failed to get me to walk into, and paid the price (or more correctly Roberts employer did, as I was told by another Journalist now fairly famous book writer that the publicity probably did wonders for their sales).
You can see a little of the run up history here,
Quantry • June 14, 2022 12:20 PM
@ SpaceLifeForm Thanks, I think. (We are also training an entire psycho-zone). Grand thought about small cell tech, and good links. I’m leery of perpetuating broadcast-anything, since it promotes the existing “hump everyone” brownie point harvest. The problem with my idea of spinning my own deep parabolic dish, is the scenario for thos who lack a line-of-sight to the service provider: Signal scatter is ultimately the source for most people, and its back to trust-everyone broadcast.
Regarding the audio artifacts during a phone call,
I get the feeling that I should also mention the potential here for
“the 10 cent glove” syndrome, mentioned by Andy Müller-Maguhn, when he found a rubber glove in his kitchen, after a campaign to radicalize him during “Operation Hotel” at an Ecuadorian Embassy (where he provided security):
you have to look at it from a cost effectiveness point of view, like that piece of plastic costs [them] ten cents, nothing, and it freaks you out three months.
h–ps://media.ccc.de/v/rc3-11512-cia_vs_wikileaks
dictionary.com:
femto-… 10^−15 …smaller
flat • June 14, 2022 12:22 PM
Who is the head of state, the “wise man who speaks little”?
xttps://www.theguardian.com/world/2022/jun/14/pope-francis-ukraine-war-provoked-russian-troops
Metaphor?
vas pup • June 14, 2022 6:22 PM
Why you may have a thinking digital twin within a decade
https://www.bbc.com/news/business-61742884
“Most of us have been told by a friend that we have a doppelganger – some stranger they passed on the
street who bore an uncanny resemblance to you.
But imagine if you could create your very own twin, an exact copy of yourself, but one that lived a
purely digital life?
We are living in an age where everything that exists in the real world is being replicated digitally –
our cities, our cars, our homes, and even ourselves.
And just like the hugely-hyped metaverse – plans for a virtual, digital world where an avatar of
yourself would walk around – digital twins have become a new, talked-about tech trend.
A digital twin is an exact replica of something in the physical world, but with a unique mission – to
help improve, or in some other way provide feedback to, the real-life version.
Initially such twins were just sophisticated 3D computer models, but artificial intelligence (AI)
combined with the internet of things – which uses sensors to connect physical things to the network –
have meant that you can now build something digitally that is constantly learning from and helping
improve the real counterpart.
…real value seen in digital twins is in healthcare.
Dassault Systemes’ Living Heart project has created an accurate virtual model of the a human heart that
can be tested and analysed, allowing surgeons to play out a series of “what if” scenarios for the organ,
using various procedures and medical devices.
The project was founded by Dr Steve Levin, who had personal reasons to want to create a digital twin.
His daughter was born with congenital heart disease, and a few year’s back, when she was in her late 20s
and at high risk of heart failure, he decided to recreate her heart in virtual reality.
Boston Children’s Hospital is now using this technology to map out real patient heart conditions, while
at Great Ormond Street hospital in London, a team of engineers is working with clinicians to test
devices that may help children with rare and difficult-to-treat heart conditions.
Experimenting on a digital heart also has the knock-on effect of cutting down on the need to test on
animals – one of the more controversial aspects of scientific research, says Severine Trouillet, global
affairs director at Dassault Systemes.
The firm now plans more digital organ twins, including the eye and even the brain.
“At some point we will all have a digital twin, so that you can go to the doctor, and we can
increasingly make preventative medicine, and make sure that every treatment is personalised,” says Ms
Trouillet.”
Clive Robinson • June 15, 2022 1:23 AM
@ SpaceLifeForm, lurker, Quantry,
Re : The Singularity is near.
Lot’s of people are getting jittery have a look at the number of religious cults talking about “the rapture” is nearly here and the nutball preppers and SHTF.
Much of the jitters is actually the “fall out” of,
1, Very Short Term Perspectives.
2, Neo-con policy coming back to roost.
3, Non resiliance in supply chains.
It’s why quite a few are “stocking up” by “panic buying” causing others to see shortages, thus respond by “panic buying”. It’s the underlying cause of the deaths of those infants due to the “infant milk formular” pathogen due to insufficient cleaning and maintenance of plant equipment. Because managment gave away to shareholders the money that should have been spebt on it, so they could get their bonuses…
But the US FDA have a very large part to play… They “rigged the requirments” for infant formular to,
1, Please the corn sugar industry of big-agro, to start infants on the path to type II diabetes.
2, Make a “closed shop” via “labeling requirments” to stop much healthier foreign imports, so please vested interests.
Expect to see a lot more of this sort of nonsense to arive on the back of world food shortages due to the psychopathic behaviours of just two men… Vladimir Putin & Xi Jinping.
Winter • June 15, 2022 1:46 AM
@Clive
Much of the jitters is actually the “fall out” of,
1, Very Short Term Perspectives.
2, Neo-con policy coming back to roost.
3, Non resiliance in supply chains.
The last decades (since Reagan won the presidency) have seen a dismantling of the USA institutions as part of the cold civil war that is raging between the neo-Union and the neo-Confederation. This decreased the already low trust level in the USA to that of a random third world country. The last health crisis has shown that this resulted in a country unable to respond to a crisis, any crisis.
I see worrying parallels between the current institutional crisis in the USA and the situation in Yugoslavia after Milosevic came into power in Serbia.
Maybe Americans do have valid reasons to have the Jitters.
Winter • June 15, 2022 1:51 AM
@vas pup
Why you may have a thinking digital twin within a decade
Thinking might be a little optimistic.
But there is definitely work done to create digital twins in the clinic. Simulating a patient to predict how they might respond to treatment in very complex situations is an active field of research.
JokingInTuva • June 15, 2022 2:16 AM
Interesting “New, Nearly-Impossible-to-Detect Linux Threat”
Clive Robinson • June 15, 2022 3:57 AM
@ JokingInTuva, ALL,
Re : ultra-stealthy Linux backdoor
It actualy is not that stealthy…
To understand why people need to get over some preconceptions / cognative bias…
Some years ago now I indicated that two papers from before Alan Turing’s seminal paper on the haltibg problem, showed that it’s not possible for a computer to be trusted. The two papers by Kurt Gödel proved that any system of logic could not be used to prove it’s self.
To quote Wikipedia[1],
“The first incompleteness theorem states that for any ω-consistent recursive axiomatic system powerful enough to describe the arithmetic of the natural numbers (for example Peano arithmetic), there are true propositions about the natural numbers that can be neither proved nor disproved from the axioms.”
I won’t go into the argument derivation but suffice it to say, you can not expect an honest answer from a computer if you ask it if it’s infected with malware. It will respond with what it is told to respond with. For those that remember back to the early days of “Rootkits” will understand one of the basic ways to do this.
Two basic ways are,
1, Modify binaries to “lie”.
2, Modofy OS to lie to the binaries.
The deeper into the system you make the changes the harder it will be to spot by examining file systems and logs.
However the computer can not hide
1, The time taken.
2, The resources usd.
By the unauthorised activities.
If these are logged by systems unknown to the attacker then the effrcts of the malware will be evident.
We’ve known this for years some may even have read about it in Cliffod Stoll’s paper “Stalking the Wily Hacker” or his book that came from it “The Cuckoo’s Egg” about his 1986 chase of a German systems invader Markus Hess[2], who was trying to gather intelligence to sell to USSR KGB. The book explains in an easily readable way how you “hide” on/in systems.
For SysAdmins who’s job it is to spot such sophisticated attacks, as I’ve indicated you can not “trust the computer”. So you have to build trust in other ways.
One of which is to run “hidden” secondary computers as “insrumentation heads” to collect information of the computers activities externally. As these will be different to those on the infected computer you can build up evidence of the malware and operator by meta-data inference.
Another way is to design a computer such that it can pick up these signitures in real time and alert the SysAdmin, it was this that formed the basis for my “Castle -v- Prison” approach which is detailed in a lot of places on this blog.
Winter • June 15, 2022 4:41 AM
@Cilve
Some years ago now I indicated that two papers from before Alan Turing’s seminal paper on the haltibg problem, showed that it’s not possible for a computer to be trusted. The two papers by Kurt Gödel proved that any system of logic could not be used to prove it’s self.
Is there actually a difference between the proofs of Gödel and Turing? And you can add Alonzo Church’s thesis into the mix.
A mathematical proof in Gödel’s proof is a computable function. Mathematical proofs can be rewritten as the halting problem. Here is one such rewrite of the Goldbach Conjecture:
ht-tps://skibinsky.com/godel-turing-and-cantor-the-math/
If you can prove that a certain computable function halts, you have proof the Goldbach conjecture is not true.
We can also go back to Russel’s problems with Set theory, the (in)famous Barber that shaves all men that do not shave themselves (and is himself a man).
Winter • June 15, 2022 4:50 AM
@Clvie
However the computer can not hide
1, The time taken.
2, The resources usd.
By the unauthorised activities.
These matters are related to Ken Thompson’s “Trusting Trust” attack (which is software only). There is a counter strategy developed by David A. Wheeler’s “Countering Trusting Trust”
Our host:
ht-tps://www.schneier.com/blog/archives/2006/01/countering_trus.html
The full theory of Countering Trusting Trust is found in David’s Thesis which is somewhere on this page:
ht-tps://dwheeler.com/trusting-trust/
I assume a hardware version of the Trusting Trust attack could be handled in the same way as the software attack if there were some hardware that was unaffected.
Leon Theremin • June 15, 2022 5:26 AM
KeyTap3 Exploit Knows What You Type; Mechanical Keyboard Eavesdropping
https://spybusters.blogspot.com/2022/05/keytap3-exploit-knows-what-you-type-via.html?m=1
ResearcherZero • June 15, 2022 6:33 AM
Major Australian retailers Kmart, Bunnings and The Good Guys are using facial recognition technology in stores, raising concerns among privacy experts.
“Kmart and Bunnings stores they visited had physical signs at the store entrances informing customers about the use of the technology, but the signs were small, inconspicuous and would have been missed by most shoppers. The collection of biometric data in such a manner may be in breach of the Privacy Act.”
https://www.choice.com.au/consumers-and-data/data-collection-and-use/how-your-data-is-used/articles/kmart-bunnings-and-the-good-guys-using-facial-recognition-technology-in-store
“The Capability”
Australia’s Department of Home Affairs has published a tender for ID document verification and face biometrics matching as components of its identity-matching services (IDMS)
https://www.tenders.gov.au/Atm/Show/d4695551-137e-4ae5-8546-33af14ce266e
lack of safeguards in the proposed law
Australia has only limited regulation of facial recognition technology. In 2017, the states, territories and federal government struck a deal to create a massive database of federal government images from driver licences and passport photos.
https://www.buzzfeed.com/hannahryan/clearview-ai-australia-police
Four recommendations have been made by the PJCIS committee, chiefly that the regime be rebuilt subject to “parliamentary oversight and reasonable, proportionate and transparent functionality”.
https://www.aph.gov.au/Parliamentary_Business/Committees/Joint/Intelligence_and_Security/Identity-Matching2019
A timeline of digital rights legislation in Australia
https://digitalrightswatch.org.au/timeline-natsecleg/
“cutting the Gordian knot”
Centralised solutions enable centralised control – and using our bodies as universal passports raises grave concerns for control over personal information and bodily autonomy.
https://www.youtube.com/watch?v=BKEM8bpYkQQ
Mr. Peed Off • June 15, 2022 11:57 AM
The US defence contractor L3Harris is in talks to take over NSO Group’s surveillance technology, in a possible deal that would give an American company control over one of the world’s most sophisticated and controversial hacking tools.
Any takeover of NSO’s hacking technology would add to L3Harris’s current suite of surveillance tools, which are already sold to US government and law enforcement clients. The company, which is based in Florida and reports about $18bn in annual sales, includes the FBI and Nato as clients.
Any potential deal faces stiff opposition from digital rights advocates and human rights groups.
https://www.theguardian.com/world/2022/jun/14/nso-group-pegasus-us-l3harris
Clive Robinson • June 15, 2022 1:33 PM
@ ResearcherZero, ALL,
Re : Australia has only limited regulation of facial recognition technology.
I can see the “Ned Kelly” look making a reappearance…
https://upload.wikimedia.org/wikipedia/commons/c/c8/Ned_Kelly_letterbox.jpg
vas pup • June 15, 2022 4:49 PM
@Winter – thank you for input.
You may like this article and usual suspects as well
Google engineer says Lamda AI system may have its own feelings
https://www.bbc.com/news/technology-61784011
vas pup • June 15, 2022 6:04 PM
Germany’s Badass New Tank Could Outmatch Every Other Tank in the World
https://www.yahoo.com/finance/news/germany-badass-tank-could-outmatch-161600880.html
“The tank, KF51, is called the “Panther,” sharing its name with a World War II-era tank.
Top-attack weapons like the Swedish NLAW and American Javelin have proven devastatingly effective in Ukraine, firing an explosive charge down into a tank’s thin roof armor. Most Western tanks are similarly vulnerable, and Panther’s top-attack defense system is the first-known system devoted to tackling top-attack weapons. It’s not clear how the new defense system works, with rumors it may involve the quadcopters moving to intercept incoming rockets and missiles.”
SpaceLifeForm • June 15, 2022 7:03 PM
@ JonKnowsNothing
I suspect planes were involved
hxtps://nitter.net/thepacketrat/status/1537160260670177284#m
Based on the number of people I’ve seen reporting COVID positive after RSA, I stand by my previous risk assessment for DefCon/Black Hat.
lurker • June 15, 2022 9:58 PM
Sorry, Mr. Dimwit here again, but
What problem are Google trying to solve by ramming Oauth2 down everyone’s throats?
Winter • June 16, 2022 1:00 AM
@vas pup
Google engineer says Lamda AI system may have its own feelings
Current AI is starting to implement Searle’s famous Chinese Room thought experiment.
ht-tps://philosophy.tamucc.edu/notes/chinese-room
Searle imagines himself in a locked room where he is given pages with Chinese writing on them. He does not know Chinese. He does not even recognize the writing as Chinese per se. To him, these are meaningless squiggles. But he also has a rule-book, written in English, which dictates just how he should group the Chinese pages he has with any additional Chinese pages he might be given. The rules in the rule-book are purely formal. They tell him that a page with squiggles of this sort should be grouped with a page with squiggles of that sort but not with squiggles of the other sort. The new groupings mean no more to Searle than the original ordering. It’s all just symbol-play, so far as he is concerned. Still, the rule-book is very good. To the Chinese-speaker reading the Searle-processed pages outside the room, whatever is in the room is being posed questions in Chinese and is answering them quite satisfactorily, also in Chinese.
AI is already doing this. We see things like GPT-3 producing convincing texts[1]. Such systems are already used to answer questions, also in Chinese. Ask a question and if the training data contains the answer, the system will print it.
But does it think or understand? It might be able to pass the Turing test. The answer depends on what you mean with “think” and “understand”. That is also the main problem with the Chinese Room thought experiment.
Now, take the old riddle of How do I explain fire, or snow, to someone who has never experienced fire, or snow? or Explain “Red” to a congenital blind person or Explain music to a congenital deaf person.
In semiology (the study of signs and semantics) this problem is called the Symbol Grounding Problem [2]. Take an encyclopedia. It contains all (or a lot) of human knowledge, but it is all words (and a few pictures). You read it, but then you have to ask yourself, what does it mean what I read?*.
To make sense of an encyclopedia, to use it, you must at some point have symbols, Red, Fire, Snow, Music that are not just words, but that you have directly experienced, without the use of words.
The solution to this problem, as far as AI is concerned, would be to equip the AI with emotions, sensors, and actuators that allows it to feel, experience, and manipulate the world. Then I would be much more willing to talk about AI being “intelligent”.
[1] Examples can be found at
ht-tps://herbertlui.net/9-examples-of-writing-with-openais-gpt-3-language-model/
More links available here:
ht-tps://findnewai.com/best-gpt-3-demos-examples/
[2] ht-tps://en.wikipedia.org/wiki/Symbol_grounding_problem
SpaceLifeForm • June 16, 2022 2:16 AM
@ Clive, Winter, ALL
“If I can design a determanistic algorithm that the observer of the output only, can not determine is determanistic, then to them it appears random”.
Two things arise from that, the first is it involves tests, the second is the test success criteria is a moving goal. That is there is no true test for “random” where as there is a partial test for determinism in that you produce the algorithm.
But can you produce the algorithm?
In theory yes in practice no.
Goes outside, observes night sky of space, finds Big Dipper …
As I mentioned before, I “know” that this magic algorithm must exist. That is the ‘theory’. But, in ‘practice’. it is not visible.
While I “know” (as do certainly most here), that the Goldbach Conjecture is definitely true, just because I “know” it is true, still leads to this strange conundrum, in that I can not provide a written proof.
I’ve studied Goldbach Conjecture for many decades. I “know” it is true.
I am quite certain that the Riemann Hypothesis is true also.
Is is not strange that we can “know” things but cannot “prove” them?
Why is math and logic so seemingly in a confusing battle at times?
But, other times, math and logic get along perfectly?
Is the reality of quantum so bizarre that we can never understand?
Clive Robinson • June 16, 2022 2:26 AM
@ SpaceLifeForm,
Re : I suspect planes were involved
Have you heard of the latest UK exports to Rwanda Africa?
I wonder what will be going with them, as health care is very definately a low priority in UK detention centers that have been privatized to the likes of “thugsRus” etc. And it is not likely health care will be any better in that part of Africa based on international monitor findings.
Of course one of the problems behind this is the “pass it along as we look the other way” of many countries. All of the imigrants hitting the UK shores in unseaworthy rubber boats and the like are “technically illegal” and “not asylum seekers” because they have been shipped through umpteen European Countries on their way to the UK. Under international law you are supposed to claim asylum at the first country you arive in not the last in a long chain (picking up god alone knows what pathogens on the way).
Though why on earth people would want to come to the UK I’ve no idea, we are rapidly going down hill, the income disparity gap is rapidly increasing and the number of people in poverty is rising significantly. Likewise the age of death gap is rising with some on the East side of London with the average age of death in the 50’s whilst just a few miles away in the West of London it’s now for some up in the 90’s. Oh and increasingly there is serious talk from Scotland, Wales and N.I. Of “breaking up the Union” to get back into the “single market” (all of which could have been avoided with a Norway++ agreement).
But why should pathogens care? They get to travel first class from their point of view, with a round the world ticket, with lots of stop overs…
Clive Robinson • June 16, 2022 2:58 AM
@ SpaceLifeForm,
Re : Connect dots.
It’s not the dots that need to be joined, but the chain of evidence, supposadly “beyond reasonable doubt”.
To be honest I do not think the prosecution can carry thst, not even close. The prosecuters opening remarks to the jury as reported already contains disprovable lies.
The real question that is going to have to be asked is,
“backup tape access”
I suspect from what has been dripped out it’s more than possible somebody “went to the tape library” and pulled a copy from there on a selected date…
As far as the information / data / source code is concerned “the system” was not “air gapped” not only were personnel in atleast two other countries able to get access, but also we know that it was “bscked up” in a more centralized system.
Thus it’s “access to data” not “access to systems” that is important and by the sounds of it the data was following the “Garden Hose Distribution” model and getting sprinkeled every which way in big arcs.
The other thing is so far there has been no evidence shown that the ex employee had anything close to sufficient “access time” to move the quantity of data claimed to have been taken.
In front of a jury with any technical competence, the evidence so far presented that has subsequently become known would not be even remotely convincing.
An interesting thought… A jury is supposed to consist of “your peers” but increasingly that is not possible. I doubt any on the jury have the level of knowledge, training, education, or experience to get even close. Likewise the defendent sounds like a high functioning individual a fair ways up “on the spectrum” and thus “not nuro-typical”. Not wishing to be nasty but the average IQ in the US is the last time I looked between 85-115 depending on which grouping you measure, and I suspect the defendants to be up around 160…
So a jury of his peers it is most definately not. Likewise I doubt the judge is much better than the jury in quite a few important respects with regards relevant “knowledge, training, or experience” and possibly education as well when it comes to such technical subjects.
Winter • June 16, 2022 3:07 AM
@SLF
As I mentioned before, I “know” that this magic algorithm must exist. That is the ‘theory’. But, in ‘practice’. it is not visible.
Actually, algorithms than spits out deterministic numbers that cannot be proven to be not random already exist. Most real numbers are not computable and there are many (almost all) that have decimal/binary expansions that show no pattern. For all we know, these numbers could be “normal”[1], that is they pass any test of randomness.
Which means, pick some irrational number for which you can calculate the expansion, π, e, √2, etc., and an offset into the sequence, and you have your perfectly deterministic sequence of perfectly random numbers.
[1] ht-tps://www2.lbl.gov/Science-Articles/Archive/pi-random.html
Clive Robinson • June 16, 2022 4:59 AM
@ Winter, SpaceLifeForm,
For all we know, these numbers could be “normal”[1], that is they pass any test of randomness.
Actually we know this to be true.
Take any number of infinite number of digits, and use it as a top row of a simple mapping function.
Then write down the next row underneath with just one digit in your chosen base changed (base 2 is the easiest).
The same in the next row but with the single digit changed being different to all preceading rows.
When you’ve done with the infinite number of single changes, go to two changes and so on.
Whilst some very very small fraction of the rows will end up with much easier determanistic algorithms to get the pattern, by far the majority by orders of magnitude will not.
But as Georg Cantor showed in 1891 with his “diagonal argument” these rows are very far from complete and thus we can show that you can not find all the determanistic algorithms that exist in theory.
But it’s also not hard to see with further consideration that every row can have an infinite number of determanistic algorithms to generate it…
So, the question then arises is there a row for which there can not be a determanistic algorithm?
I think you will find the answer is no. Therefore there will always be a “generator function”
But lets go one step further and say can we describe a universal generator function?
The answer is “yes” the mapping function is a “lookup table” as well indexed by the row address abd individual base digit column number.
Knowing this how do we then “define random”?
The simple answer is we can not except in non testable terms…
Winter • June 16, 2022 5:36 AM
@Clive
So, the question then arises is there a row for which there can not be a determanistic algorithm?
A “deterministic algorithm” is here assumed to be an algorithm whose “code” is much smaller than the resulting “normal”/”random” sequence. Then this problem comes down to the Kolmogorov complexity argument.
In terms of Kolmogorov complexity, the number of algorithms is a countable set.
There are ≤ 2^m algorithms that fit in m bits. There are 2^N possible sequences of N bits. If m<<N, which is how we define a "deterministic algorithm" that generates a pseudo-random sequence [2], then it follows that most sequences (≥ 2^N – 2^m) cannot be generated by a deterministic algorithm.
[2] If m ~ N, then the algorithm itself becomes a random, unpredictable, sequence.
Winter • June 16, 2022 6:35 AM
About AI becoming “Intelligent” instead of “Stupid”
We all know advertisement like this, that shows how small modern IC are:
ht-tps://pbs.twimg.com/media/B-U3Z0nCYAAlTDk.jpg
[alt][Ants carrying electronic chips]
But in reality, these ants are way smarter than the AI’s creating by wiring up millions of these ICs.
Look to insects if you want to build tiny AI robots that are actually smart
ht-tps://www.theregister.com/2022/06/16/insects_ai_robots/
“We argue that inspiration from insect intelligence represents an important alternative route to achieving artificial intelligence in small, mobile robots,” they wrote. “If we succeed in harnessing insect-inspired AI, small robots will be able to tackle difficult tasks while staying within their limited computational and memory budget.”
I would say, a small insect navigating a random garden or forest to forage, avoid predators, find a mate, and find a place to lay eggs, already improves massively on a Tesla auto-piloting at high speed into some large structure because it did not recognize that kind of structure.
Clive Robinson • June 16, 2022 6:38 AM
@ Winter,
A “deterministic algorithm” is here assumed to be an algorithm whose “code” is much smaller than the resulting “normal”/”random” sequence.
I make no such assumption and with good reason, we are talking infinite sequences, therefore any repeating pattern will match “whose code is much smaller”.
There are an uncountable number of sequences that do not repeate that we know can be found by a short algorithm, through to those with algorithms that are longer than we can determin, though certainly shirter than the sequence the generator will produce.
As a rule of thumb, the longer a sequence is the more likely it is to be “compresable” arguably there may be no infinite sequence that can not be compressed in some way, I actually suspect that is the case.
Kolmogorov complexity kind of applies to short numbers in a finite mapping set, all of which can be found by a minimal change to another member of the set. Thus the complexity is realy one of “distance” between any two members of the set and finding the most optimal path and order of the many there are (many many times the number of members in the set).
Winter • June 16, 2022 6:58 AM
@Clive
I make no such assumption and with good reason, we are talking infinite sequences, therefore any repeating pattern will match “whose code is much smaller”.
A main (trivial) result in Kolmogorov complexity is that for any finite size set of sequences of length N (bits), 2^N sequences, there are only O(2^m) sequences that can be produced by algorithms of size m (bits) or less.
A repeating pattern of size l can be encoded in an algorithm of size O(l) (bits). The fraction of sequences of length N (bits) with repeating patterns of any kind with range l<<N (say l ≤ √N) becomes vanishing small for large N.
I do not see the problem. If there are 2^N sequences, and only O(2^m) possible programs of size m << N, how would it even be possible that these O(2^m) programs could generate most of these 2^N sequences?
There have to be many sequences that cannot be generated by programs significantly smaller than the sequence itself.
Winter • June 16, 2022 9:17 AM
Russian spy caught seeking to infiltrate ICC: Netherlands
The Dutch authorities uncover a Russian military spy posing as a Brazilian intern to infiltrate the Hague-based international war crimes court.
ht-tps://www.aljazeera.com/news/2022/6/16/russian-spy-caught-seeking-to-infiltrate-icc-netherlands
Sergey Vladimirovich Cherkasov created an elaborate cover story dating back years to attempt to enter the Netherlands as a Brazilian national for an internship at the Hague-based ICC in April, the agency’s head told the Reuters news agency on Thursday.
“This was a long-term, multi-year GRU operation that cost a lot of time, energy and money,” said Dutch intelligence agency chief Erik Akerboom, using the acronym for Russia’s military intelligence service.
…
The Dutch agency said it had taken the unusual step of releasing detailed information on the case to expose the workings of Russian intelligence and the threat to other international institutions.
It distributed a four-page document outlining what it said was Cherkasov’s invented cover story, which included a supposed troubled family history and details from a club where he liked to listen to electronic trance music and his favourite restaurant in Brasilia where he would eat cheap brown bean stew.
Winter • June 16, 2022 9:30 AM
In an unusual attempt of the Dutch intelligence community, the AIVD, to inform the public about what happens, the elaborate cover story of the spy Sergey Vladimirovich Cherkasov who tried to infiltrate the ICC was published in full (English version [1]):
ht-tps://www.aivd.nl/binaries/aivd_nl/documenten/publicaties/2022/06/16/bijlage-dekmantel-van-russische-inlichtingenofficier/Bijlage+met+dekmantel+Russische+inlichtingenofficier-Engelse+versie.pdf
See previous comment:
https://www.schneier.com/blog/archives/2022/06/friday-squid-blogging-squid-changes-color-from-black-to-transparent.html/#comment-406265
[1] For Dutch and Portuguese see:
ht-tps://www.aivd.nl/documenten/publicaties/2022/06/16/bijlage-dekmantel-van-russische-inlichtingenofficier
Clive Robinson • June 16, 2022 12:04 PM
@ vas pup,
Re : USB-C “as a standard”
As I mentioned one of the problems I indicated with USB-C is the combined communications and charging.
What I did not make clear is that part of that “communications” is not “user data etc” but actually what voltage the charger should send to the device being charged/powered and current limitations.
A lot of people might say “So what?” untill they find their device will not charge or power from a working USB-C power source.
Well it’s becoming more of an issue for some people who travel.
A short You-tube vide showing this for some quite ordinary devices, and as the man says “standards should be standard” 😉
And it’s realy annoying when they are not,
https://m.youtube.com/watch?v=f0lbKIA34gQ
Oh the other issue with USB these days is politely refered to as “the wetted contacts issue” basically all meyals oxidize or form non-conductive layers with the atmosphere (yes even gold). For a reliable electrical connection the insulating layer needs to be “broken through” so the contacts in connectors are,
1, Often made of soft metal.
2, Often designed to scrape.
3, Are increasingly getting smaller.
The result is “reliability is lower than it has been in the past.
Some quite expensive connectors are only guarenteed for as few as “fifty operations” which does not sound very reliable… But the issue is they degrade with time, which means the “signalling” used for “Power Distribution” gets flaky faster than your multimeter probes will show. Importantly some cables will work in some devices and not in others so diagnosing problems is going to require people to be “Smarter than the average bear”.
Quantry • June 16, 2022 12:38 PM
@ Clive Robinson
I loosely get that you saying that “panic buying” and related infant deaths are at least in part or indirectly caused by “religious cults”…
An old WHO code regarding Child Feeding seems to blame marketing sources also:
h–ps://apps.who.int/iris/handle/10665/40382
And for the reasoning:
h–ps://www.unicef.org/mena/topics/breastfeeding
plus a bunch of good sources like
h–ps://www.asklenore.info/breastfeeding/toxins.shtml
(My “go farming / grow a garden” advice is based on evidence of what I see in society. And I am less faithful than anyone who is able to believe “everything came from a huge explosion of nothing in a completely empty vaccuum”.)
I agree that frenzied thinking from all sources like “weforum” to “to shape global, regional and industry agendas” (by extreme propaganda it seems), is pervasive.
Cheers.
lurker • June 16, 2022 1:11 PM
@Clive, re USB-C contacts
The first time I plugged in a USB-C connector I was astonished how short the “travel” was. That is from first feeling contact mechanical resistance to fully home is typically less than 2mm. That’s not much contact area on those narrow pins for 5 amps charging, yes even with 4 in parallel it’s still 1.25 amps per pin, and we all know what DC does to sliding contacts.
John • June 16, 2022 1:15 PM
@Clive,
Hmm….
Remember all those tin problems. Oxides and whiskers!!
‘re-seating’ computer cards.
Multiple grounds and power fingers.
original USB was OK. USB-C is a standards disaster.
I am surprised that Europe has apparently not figured that out.
John
vas pup • June 16, 2022 3:53 PM
Taiwan shows off latest home-made armored vehicle
https://news.yahoo.com/taiwan-shows-off-latest-home-063744024.html
“The eight-wheeled CM-34, which entered service in 2019, is armed with the Mk44 Bushmaster 30mm chain gun, made by Northrop Grumman and with an effective firing range of 3 km (1.9 miles). It is designed to be highly mobile, operating in all weathers.”
vas pup • June 16, 2022 4:11 PM
Democrat senators led by Elizabeth Warren want to ban brokers from trading people’s health and location data
https://news.yahoo.com/democrat-senators-led-elizabeth-warren-095645680.html
“A bill introduced Wednesday by Warren, cosponsored by Bernie Sanders and several other Democrats, would ban data brokers from selling or transferring individuals’ location data and health data, according to a statement on Warren’s website.
“Data brokers profit from the location data of millions of people, posing serious risks to Americans everywhere by selling their most private information,” Warren said.”
Agree.
Q: Do we need prevent access to PII of US residents by customer service representatives located outside US in particular working for banks, processing financial transactions involved PII?
I guess that is something to think about for Legislators of both major parties.
SpaceLifeForm • June 16, 2022 6:07 PM
@ Winter, Clive
Actually, algorithms that spits out deterministic numbers that cannot be proven to be not random already exist.
Interesting double negative.
There are two algorithms in play here, the generator, and the observer prover.
But yet, it appears to this observer, that the contrapositive is False. I.E., this statement is False:
Algorithms that spits out random numbers that can be proven to be deterministic already exist.
That may be True, but does it really exist?
Is there some magic algorithm that can prove this? Maybe. But maybe it is not visible.
And, what does that statement tell us?
The problem is that if there is a ‘proof’, then the observer algorithm has already figured out what the generator algorithm is doing, and therefore it is deterministic, and therefore cannot be generating random. Therefore the premise must be incorrect.
I’m not trying to put words in your mouth, but if the contrapostive is False, then the original statement must also be False, right?
It’s just logical, right?
hxtps://en.m.wikipedia.org/wiki/Contraposition
SpaceLifeForm • June 16, 2022 6:42 PM
@ name.withheld.for.obvious.reasons
It appears that USAG Garland has connected the dots.
lurker • June 16, 2022 8:06 PM
@vas pup
Q: Do we need prevent access to PII of non-US residents by customer service representatives located inside US?
The answer to that will probably depend on a widely held US assumption that anyone coming into the US[1] to do business must be prepared to give up at least some PII.
Can the Legislators of either major (US) party apply the same logic to your question? The US is one of the few countries that has goose-sauce and gander-sauce, different recipes.
[1] We already know that if any network traffic passes through a US owned node the business is deemed to have occurred in the US.
SpaceLifeForm • June 17, 2022 12:47 AM
@ Winter, Clive
There have to be many sequences that cannot be generated by programs significantly smaller than the sequence itself.
While this is True, as a practical manner, does it really matter when it comes to Cryptography and Security?
There are plenty of other sequences that are usable as a OTP. We do not need to address the infinities involved.
So, back to
Which means, pick some irrational number for which you can calculate the expansion, π, e, √2, etc., and an offset into the sequence, and you have your perfectly deterministic sequence of perfectly random numbers.
Now, think about multiple offsets, using the irrational bit sequences, with the bits XORed together to create the OTP, on the fly.
As long as I can securely tell you the offsets, and which irrational bit sequences are being used, then I can create the OTP, encrypt the plaintext, put it somewhere, and you can recreate the OTP and decrypt the ciphertext.
The key to this method is whether we can agree upon the rules. But, if we both have, say 1Mbits of π, e, √2, etc. in our program, then as long as I can tell you, here are the offsets to use into the irrational bit sequences, how many offsets there are, and which irrational bit sequence is to be used by each offset, then you can recreate the OTP.
I do not have to securely transmit the entire OTP via courier.
The key (the offsets and the irrational bit sequence ids) can be much smaller than the OTP. We do not need to address all possible OTPs. There are so many it does not matter. Remember, this is One Time.
If the key is say 256 bits, generated at Random, the odds of one re-using a OTP is vanishingly small.
The main problem is that the crypto and the comms must be separated, but also, I must be able to securely send you the key. I will need to get the ciphertext off of the offline machine (to transmit it somewhere), and also get the key off of the offline machine (to transmit it to you). Those two steps need to be separate, and go thru different channels. This is the scenario that nearly no one will ever want to entertain.
Convenience, Security. Pick one.
Maybe this is why Numbers Stations exist.
SpaceLifeForm • June 17, 2022 1:10 AM
Russian fires
I was kind of expecting this a couple of weeks ago.
Should be visible via satellite. But here are some ground level videos.
I imagine this one will be tricky.
hxtps://nitter.net/igorsushko/status/1537478732839002112#m
Winter • June 17, 2022 1:20 AM
@SLF
There are two algorithms in play here, the generator, and the observer prover.
Let’s make it straightforward:
First a useless result:
Infinite sequences can be thought of as the decimal expansion of real numbers and programs as natural numbers. There are more Real Numbers than natural numebrs (Cantor). Which means there are Real numbers that are not computable.
This is a useless result as every actual number used will have a finite expansion and will therefore be a rational number.
But.
A sequence of length N is defined as non-random if there is no program shorter than the sequence (actually, there is some fixed overhead to take into account) that can generate the sequence and halt[1]. This is easy to prove for any finite sequence: Just enumerate and run all programs shorter than the sequence (+overhead). But this becomes unpractical pretty soon. So, it is necessary to limit the size of the programs to some value m < N for practical lengths N.
For every limit in the size of programs ≤ m (bits), there is a limit in the number of sequences that can be generated of 2^O(m) ≤ 2^m. The number of sequences with length N > m is 2^N > 2^O(m). Therefore, there must be sequences of length N that cannot be generated by a program of size m.
It follows from our “definition” of randomness that if N=O(m), then selecting a program is just as random as selecting the sequences themselves so the programs are the random sequences. But the programs are only useful if the length of the sequences, N, is (much) larger than the length of the programs, m, to generate them, N >> m.
If N >> m the number of sequences that can be generated by the programs, 2^O(m), is a vanishingly small fraction of the number of possible sequences, 2^N.
What is problematic in this argument?
[1] More than you probably want to know about Kolmogorov complexity can be found at:
ht-tps://eprints.illc.uva.nl/id/eprint/996/1/CT-1988-01.text.pdf
Winter • June 17, 2022 1:25 AM
@Clive et al.
Re : USB-C “as a standard”
Apple has used USB-C (Thunderbolt) compatible connectors in iMacs and Macbooks for years, even with higher currents than specified in USB-C. There have been Android phones using USB-C for years.
Where are the problems?
Winter • June 17, 2022 1:51 AM
@SLF
Now, think about multiple offsets, using the irrational bit sequences, with the bits XORed together to create the OTP, on the fly.
That is security by obscurity. It is possible to compile a rainbow table of “known irrational numbers” with generators. Then the only unknown will be the offset, which will be a ~48 bit number.
But the real bummer is that these generators are basically PRNGs[1], but then very inefficient ones. Why not simply use an efficient PRNG?
[1] ht-tps://ieeexplore.ieee.org/abstract/document/8367171
Clive Robinson • June 17, 2022 8:57 AM
@ SpaceLifeForm,
I do not have to securely transmit the entire OTP via courier.
If the KeyMat is of less size than the message thrn it can not be an OTP, because it will fail the important unicity distance.
I do not know why it does not get mentioned more often, because it’s actually quite important and lifts the OTP above all other crypto systems by a very long way.
As I point out frequently the security of the OTP is not that you can not find the plaintext from the ciphertext but,
“All messages of equal length or less are equiprobable”
That’s very far from true for nearly every other crypto system.
Why is this important?
Because of “DENIABILITY” or if you prefere “BETRAYAL” by the second party to a third party.
Look at it this way,
1, Alice invites Bob over and he spends the afternoon copying out the OTP by his hand onto forms he brought over.
2, At some later date Alice sends Bob an enciphered message saying something like “attack on twenty fifth”.
3, Eve gets a copy of Alices sent ciphertext.
4, Eve captures Bob or offers Bob a deal so that Alice can be dragged into court.
5, Bob hands over his copy of the ciphertext and his copy of the OTP written in his own hand on his own paper with the grid printed from his own printer.
Provided Alice has been carefull not to tie any message content she might have sent or received to her observable activities what evidence has Eve got against Alice?
If she was smart she would have generated innocent messages and made fake KeyMat for them which do tie up with her observable activities.
The result is infront of an independent arbiter all Eve has is Bob’s self incrimination…
You can not get any other cipher system that uses shorter than message keys to do that, or anywhere as near as easily.
Now… As I’ve pointed out before you should never send ciphertext but fairly ordinary plaintext.
What you do is use “standard / stock phrases” such as salutations to send just a few bits at a time.
So {Hi, Howdy, Watcher, What Ho,} can be used to send a two bit number. Similar can be done with such stock phrases in emails and the like.
The number then becomes the index into a “Code Book” of messages similar to those “Messages to our friends” sent by the BBC during WWII (effectively a short message numbers station).
The problem is that if the number gets used for the same action repeatedly then Eve can build up a correlation between what Alice sends to Bob and what activities happen.
If however the binary number is encrypted with an OTP not only is the corelation no better than random, Alice has deniability…
Not “perfect” security wise but nothing ever is, but it should limit issues if you are dealing with “impartial” judges and juries.
lurker • June 17, 2022 1:32 PM
@Winter, “Where are the problems?”
The problems include modern users expectations that early failure is “normal” in the throwaway society, and feature creep that persuades users to upgrade before they are forced by hardware failure.
Winter • June 17, 2022 4:22 PM
@lurker
The problems include modern users expectations that early failure is “normal” in the throwaway society
Standardized USB-C replaces a zoo of incompatible plugs that are not better. This zoo leads to large amounts of chargers and cables ending into the waste.
Not standardizing seems to me the worse way to proceed.
SpaceLifeForm • June 18, 2022 6:08 PM
@ Winter
That is security by obscurity.
Yes it is. By design.
It is possible to compile a rainbow table of “known irrational numbers” with generators.
It is not rainbow, the “known irrational numbers” are clearly hardcoded in the software. This is also by design.
Then the only unknown will be the offset, which will be a ~48 bit number.
Reread what I said. Multiple offsets.
In my example, I mentioned 256 bits.
Consider 16 offsets each 16 bits wide.
It could be more offsets, or wider offsets.
The more offsets, the slower the crypto will be. That is a given. But, does Eve know what the offsets are, which irrational bit sequences are being used for each offset, and how many and which offsets are actually being used?
P.S. Stop doing Preview.
Clive Robinson • June 18, 2022 7:38 PM
@ Winter,
With regards,
There are more Real Numbers than natural numebrs (Cantor).
Yes but remember the “proof” is actually a “generator function” in it’s own right.
Which means there are Real numbers that are not computable.
As I’ve indicated the proof is a generator which implies the opposite of that statment.
The way out would be to provide a proof of these “not computable” numbers, with a proof that can be proved to not be a generator.
Oh and “appealing to the audiance” by claiming infinite or unbounded, won’t help because there is no reason why the generator computation can not run infinitely or unbounded.
All that is required is that it produces output as it runs.
So one argument would be a two part algorithm,
1, A simple generator algorirhm.
2, That selects the rows and columns from a –potentially infinite– map.
Which is what the Cantor argument is as a generator.
SpaceLifeForm • June 18, 2022 7:52 PM
@ Clive
Reread what I said.
Those two steps need to be separate, and go thru different channels.
This is to prevent Eve from finding any correlation.
Add noise KeyMat.
Add noise ciphertext.
Do you really think that Eve can figure this out? That traffic analysis will work?
Serious question. I do not think that Eve can do the correlation. Convince me that I am wrong.
I am serious. Convince me that Eve can correlate the traffic and figure out who Alice and Bob are.
Serious. Tell me why that I am wrong.
I do not believe you can change my mind, and, with all due respect, I ask that you think outside the box.
MarkH • June 18, 2022 10:03 PM
Re: Math
Hoping to shed some light (and not dark!) on matters discussed above:
1) I’ve read the most reals are uncomputable, but haven’t found a proof or citation thereof. If anybody can point to one, I’d appreciate it.
2) Where Clive wrote “generator,” my interpretation is that he meant what is usually called a constructive proof, which essentially argues that “we know a mathematical entity with the certain specified characteristics must exist, because the following procedure shows how to construct an instance thereof.”
Constructive proofs are distinct from non-constructive proofs of existence, which show by logic that a mathematical entity with the certain specified characteristics must exist, without providing any example. Some mathematicians were quite offended by this notion when it was introduced around the turn of the 20th century. From such proofs, it is known that entities must exist within certain categories, even though no such thing has ever yet been discovered.
MarkH • June 18, 2022 10:04 PM
continued:
3) According to my very superficial understanding of Turing’s proof, it does not offer a construction of an uncomputable number, but rather shows that the computation will stop yielding digits at some point.
4) There are many excellent reasons to mistrust computers, but I think it’s a mistaken interpretation that proofs in mathematical logic “showed that it’s not possible for a computer to be trusted.”
For example, that there is no single algorithm for the halting problem doesn’t preclude that particular programs — or even an infinite family of programs — can be proved to complete after a finite number of steps.
JonKnowsNothing • June 18, 2022 10:31 PM
Oh well – no luck on the compost heap and no luck on replanting in multiple buckets.
Perhaps there will be another time to review the security implications of spending ($59B-$103B * N-Areas * M-Geographic Regions * Y-Population) in economic social health assistance for the near and mid term.
JonKnowsNothing • June 18, 2022 10:42 PM
@MarkH
re: There are many excellent reasons to mistrust computers…
Recently I showed someone how computers lie.
I had them take 3SqRt and 7SqRt and then X2 the results. (1)
The result auto-adjusted back to integers 3 and 7.
They didn’t quite understand why rounding the number was wrong.
I used the analogy of Interest Rates and asked if they would be happy if their 6.457 interest rate was rounded up?
- If it was their mortgage rate they wouldn’t like it at all
- If it was their pension fund they would be very happy
Computers lie but sometimes we like the lies enough to not be bothered by them.
===
1) My maths formula might not covert ymmv
Clive Robinson • June 19, 2022 12:13 AM
@ SpaceLifeForm,
Re : OTP and unicity distance.
I think you and I are discussing two entirely seperate issues.
I am talking about deniability under a state of betrayal and you are talking about some form of “Key Managment(KeyMan) that involves “Key Generation”(KeyGen) and “Key Distribution”(KeyDist).
That is the system you describe sounds like a “directed walk through a map” where the first party sends the second the path way-point coordinates of a “shared secret” map.
Which sounds the equivalent of a random block cipher being used in ECB mode driven by a counter that is frequently updated by the coordinates giving plaintext to the block cipher and the ciphertext output from the random block cipher being used as “the key-stream” for a subsequent stream cipher.
Under the “betrayal” condition,
A, Eve would have seen and recorded the ciphertext from Alice to Bob when it occured.
B, Bob would subsequently betray all he has to Eve.
That is the second party in the ciphertext communication between Alice and Bob –that Eve has seen and recorded– would reveal to Eve,
1, The “Shared Secret” map.
2, All the path coordinates.
3, The method used to convert the path coordinates to the Key Stream numbers.
4, The decrypted plaintext.
Eve would be able to verify the plaintext by doing the same steps as Bob.
The two important things to note are,
X, At some point unless the map is of infinite size the coordinates sent by Alice to Bob will “cross over” causing “KeyMat reuse”
Y, The KeyMat generated by the coordinates with the ciphertext must produce valid plaintext.
The probability of Alice making a second “shared secret” map that always makes valid but innocent plaintext goes down very rapidly with the amount of stream cipher KeyMat generated. For instance with both DES and AES used as the map it’s just a couple of blocks when they are used in stream generator systems. Worse Alices ability to find such a maping is inversely related to the power of the block length, and becomes effectively impossible after just a few bits (as the unicity distance predicts).
With a genuine OTP where there is no dependence in anyway between the individual stream cipher bits therefore the unicity distance is unbounded and Alice can make as many faux OTP’s as she wants with next to know difficulty.
The important point being that despite his betrayal of Alice to Eve, in every way he can, Bob actually gives nothing to Eve when a proper OTP is used. Because there is nothing to tie the plaintext Bob claims to be from the ciphertext Alice sent. Unless Eve observes and records some correlating behaviours by Alice Eve has nothing to present to an impartial judge and jury.
Clive Robinson • June 19, 2022 2:13 AM
@ MarkH,
1) I’ve read the most reals are uncomputable, but haven’t found a proof or citation thereof. If anybody can point to one, I’d appreciate it.
The “simple logic” argument that an infinate sequence that does not repeate and has no discernable relationship between the digits is “uncomputable”, is in no way a proof.
Because the ratio between any two or more reals will most probably produce another real, and computing a ratio is not that difficult, even though it might require considerable resources. A crude but inificient way is division by successive subtraction and shifting, we get taught at school, which can be speeded up by doing partial subtractions from most significant digits downwards using simple carries.
So I suspect there is “no proof” that reals are uncomputable, though some almost certainly do fall into the “uknown” catagory.
That is it is probably not too difficult for Alice to come up with a “generator” algorithm that observer Bob can not find a “prover” algorithm to match except over very short sequence lengths (for instance there is a section of pi where there is a simple sequence match for ten or twenty digits). In fact it’s probably easier to find a generator that a prover can not be found for, than it is to come up with a short algorithm that uses finite resources to come up with an infinate length sequence that does not repeate.
Speaking of “generators”,
2) Where Clive wrote “generator,” my interpretation is that he meant what is usually called a constructive proof, which essentially argues that “we know a mathematical entity with the certain specified characteristics must exist, because the following procedure shows how to construct an instance thereof.”
Sort of but a bit more constrained in that it must pe practical, both in terms of physical resources –for storage and computation– and in time to start producing usable output.
That is a finite amount of storage of state, that gets “seeded” in a reasonable time and an algorithm that produces output almost immediately from the state, and a state update algorithm that is equally as fast.
Which brings us onto,
3) According to my very superficial understanding of Turing’s proof, it does not offer a construction of an uncomputable number, but rather shows that the computation will stop yielding digits at some point.
Err no. It shows you can not know if it will stop yielding digits or not.
A superficial examination of a Turing Universal Engine as a State Machine shows it can work with an infinite length tape as it only has a notion of “relative position and range” with respect to the tape reading head and a unique indicator pattern it can write to the tape at any point to use as a start position.
Which brings us onto,
4) There are many excellent reasons to mistrust computers, but I think it’s a mistaken interpretation that proofs in mathematical logic “showed that it’s not possible for a computer to be trusted.”
There you and I see things differently, and rather depends on your notion of “trust”.
What can be said is that a single Turing Universal engine is a slave to what is on it’s tape. That is what it does with regards head movment and what it outputs are entirely dependent on what is on the tape. It has no notion of time passing or if the contents of the tape are being changed in some way (noise, degradation, malicious third party) Importantly it can not therefore telly you if it’s tape has on it what you think it has on it, only what is on the tape tells it to tell you.
We know that from the various “liar” riddles that certain rules of logic have to hold inviolate. For instance behaviour has to be consistant so that passing your question through two sentinels one that is honest and one that is dishonest in any order will always produce a false result.
However taking that a step forward where a sentinel is programable becomes mor difficult. The fact that it can be shown that a system of logic can not adiquately describe it’s self tells us that all programable systems with mutable memory can not be inviolate therefor they can not be trusted.
That is they do what the tape tells them to do and because of that they can not inherently tell if what is on the tape is what should be on the tape or not. The very “universitality” of the Turing engine / state machine is it’s own downfall when it comes to “trust” because there can be no “root of trust” to be built upon.
So no “inviolate” root of trust and no means to check it is “inviolate” means it can not be trusted.
How you go about resolving that is a “turtles all the way down” problem with Universal Turing Engines, which brings us back to a variation on the “halting problem” as well as the “undecidability” problem.
Thus you can not tell if the sentinel is a liar or not because you can not know if it is Cretan or not (Epimenides paradox).
But even asking one sentinel what the other sentinel has to say has to be treated with care. As you can extend the notion of “All Cretan’s are liars” with the sybling or liar cycle paradox,
Alice says Bob tells the truth
Bob says Alice tells lies
All you can realy deduce is one is a liar but not which.
But what of,
Alice says Bob tells the truth
Bob says Alice tells the truth
Or,
Alice says Bob tells lies
Bob says Alice tells lies
What can you reason about Bob and Alice?
I could go one but…
Winter • June 19, 2022 4:26 AM
@MarkH, Clive
1) I’ve read the most reals are uncomputable, but haven’t found a proof or citation thereof. If anybody can point to one, I’d appreciate it.
ht-tps://math.hmc.edu/funfacts/computability-of-real-numbers/
tl;dr
Algorithms are countable, reals are not countable. There are more reals than programs, hence, some reals (infinitely many) have no program to print them out
SpaceLifeForm • June 19, 2022 4:39 AM
@ Clive
Re : OTP and unicity distance.
I think you and I are discussing two entirely seperate issues.
No, I don’t think so. I think we are just looking at the problem from different angles.
The betrayal problem exists independently of whether Eve is around to observe. Bob can always betray Alice. Or vice versa. We must conclude that there are reasons that Alice and Bob are communicating in a secure manner. If they don’t trust each other, then why are they communicating in the first place?
Unless Eve observes and records some correlating behaviours by Alice Eve has nothing to present to an impartial judge and jury.
Yep. Observation and Correlation.
While we must assume Eve can Observe, can she Correlate?
And can Eve determine that Alice and Bob are communicating at all?
My entire thought process on this is simple. Make it difficult for Eve.
Society should not assume a default of Panopticon.
SpaceLifeForm • June 19, 2022 5:31 AM
@ Clive
The fact that it can be shown that a system of logic can not adiquately describe it’s self tells us that all programable systems with mutable memory can not be inviolate therefor they can not be trusted.
Cosmic ray strikes again.
Winter • June 19, 2022 5:35 AM
@Clive
The fact that it can be shown that a system of logic can not adiquately describe it’s self tells us that all programable systems with mutable memory can not be inviolate therefor they can not be trusted.
I would follow Descartes and not trust myself, nor my senses, nor my thinking.
There is only one unshakable truth: Cogito ergo sum.
Winter • June 19, 2022 7:27 AM
@MarkH, Clive
1) I’ve read the most reals are uncomputable, but haven’t found a proof or citation thereof. If anybody can point to one, I’d appreciate it.
(continued)
A proof that most finite strings are incompressible and, therefore cannot be generated by a smaller program is given in Wikipedia:
ht-tps://en.wikipedia.org/wiki/Kolmogorov_complexity#Compression
A string s is compressible by a number c if it has a description whose length does not exceed |s| − c bits. This is equivalent to saying that K(s) ≤ |s| − c. Otherwise, s is incompressible by c. A string incompressible by 1 is said to be simply incompressible – by the pigeonhole principle, which applies because every compressed string maps to only one uncompressed string, incompressible strings must exist, since there are 2^n bit strings of length n, but only 2^(n − 1) shorter strings, that is, strings of length less than n, (i.e. with length 0, 1, …, n − 1).
In other words, the number of possible PRNGs produced strings is only a vanishingly small fraction of all possible OTPs.
Clive Robinson • June 19, 2022 7:27 AM
@ Winter,
Algorithms are countable, reals are not countable.
Be carefull you are not comparing Apples with Apples[1]. The link you gave was not what it could have been[2].
There are more reals than programs, hence, some reals (infinitely many) have no program to print them out
There is an infinite number of reals between zero and one, likewise between every positive and negative countable number (interger) to the infinities.
But a simple algorithm of “Real plus Real” gives a whole infinity of infity of algorithms. Likewise “Real minus Real” and the same with multiplication and division.
Thus algorithms taken as their numeric form, far outnumber the number of reals.
So we are looking for some more equatable or equivalent notion to seek a proof.
[1] From René Descartes’s OF THE PRINCIPLES OF HUMAN KNOWLEDGE : Part 1 § 5
Why we may also doubt of mathematical demonstrations.
“We will also doubt of the other things we have before held as most certain, even of the demonstrations of mathematics, and of their principles which we have hitherto deemed self-evident; in the first place, because we have sometimes seen men fall into error in such matters, and admit as absolutely certain and self evident what to us appeared false…”
[2] From the link,
“We first show that the set of all possible computer programs is a countable set. Why? Every computer program is a finite string of a finite set of symbols. So the set of all computer programs of length N is finite.”
Actually not true. The underlying assumption is all “fully determanistic programs incapable of modification” of which only a very small subset of the programs written belong to. In part it’s one of the reasons Alan Turing pushed so hard for having a random number generator put in the Manchester Computer. But since then we have had programs that can rewrite themselves.
Interestingly there is a subset of sequence generators that generate sequences far in excess of the program length by thousands of millions of times, Pi being one the square root of two being another. Of further interest is some can be written in a way that does not require very much “active state” thus in theory could go on for ever with a Turing Engine that has an infinite tape available.
And that is where their proof fails completely, because they alow an infinity of infinities for one (numbers) but not the other (algorithms) so it is no proof at all.
Clive Robinson • June 19, 2022 8:00 AM
@ SpaceLifeForm,
First lets give an answer to,
We must conclude that there are reasons that Alice and Bob are communicating in a secure manner. If they don’t trust each other, then why are they communicating in the first place?
Just to get it out of the way, because it’s not realy relevant.
1, Assume they did trust each other.
2, Assume they are working in a hostile environment where unlimited duress can be applied via say the fact that Bob’s brother is going to be executed or similar.
There are other variations but it’s best to assume you will always be betrayed when somebody sees a benifit they can? exploit in some way.
To assume otherwise would be foolish, but sadly very very human (which is why thumb screws on others works better than thumb screws on those being coerced).
So having “eaten the vegtables onto the meat”…
[C]an Eve determine that Alice and Bob are communicating at all?
If “collect it all” is in place on the communicating medium the answer is probably “yes” but “Does Eve see it unless it is drawn to her attention in some way?”
You appear to think the same,
While we must assume Eve can Observe, can she Correlate?
But actually it goes deeper than this. You may remember that for some time I’ve said Tor has failings. But one I consider most grevious is that the end points were “leaf nodes” and not actually part of the network proper. So the user or server traffic was visable independent of other traffic and link stuffing etc, so was available to Eve to use via “Collect it all” and cross correlate the end points with each other via various types of “Traffic Analysis”.
So we need a better “Network on the network” to eliminate these issues. Something we do not currently have[1]. But it’s not just Tor all the supposed “secure” communications apps etc actually are not secure. At best they are just an easily avoidable bit of “security by obscurity”.
Which brings me to,
My entire thought process on this is simple. Make it difficult for Eve.
Well try finding modern communications that are anthing but “security by obscurity”…
I think you know what I’m hinting at. Especially when you say,
Society should not assume a default of Panopticon.
Sadly I realy do not see us doing otherwise.
[1] Personally I would not use Tor to dowbload old newspaper articles, let alone anything else.
Winter • June 19, 2022 8:11 AM
@Clive
Thus algorithms taken as their numeric form, far outnumber the number of reals.
You quoted Cantor as proving there are more reals than natural numbers.
Programs can always be interpreted as natural numbers. Programs are countable, reals are not. There are more reals as programs, even though both are infinite.
Clive Robinson • June 19, 2022 9:19 AM
@ Winter, ALL
First off a philosophical question for everyone,
“What is the difference as far as printing is concerned, of an integer of infinite length and a real of infinite length?”
To which the answer is
“One may have a decimal point and the other does not need a decimal point.
You can also abstract the point out of the string by replacing it with a second “position” number that is just a “string of digits”.
So in fact, “strings of digits” of infinite length can be any type of number in any kind of base. The fact that the point can be in an infinite number of places does not actually change “the string of digits” in any way.
Something people somehow forget when they are looking for “easy explanations” and faux proofs.
So from a string perspective there are as many countable numbers as there are uncountable numbers, and this is important when you start talking about computer programs.
Because what does,
Program = Algorithm + Data
Realy mean?
The reality of most programs is
Program = Algorithm pointers + Data pointers
Or the way most programs are written these days,
Program = pointers
Thus if we have,
Ptr1 = pointer to structure 1
Ptr2 = pointer to structure 2
Where the structures are
|type|pointer to data|
You could write a program that when loded simply pushes Ptr1 onto the stack followed by Ptr2 onto the stack.
Where Ptr2 points to structure 2 where the type is “Code” and the data pointer points to a string “Print TOS”
And Ptr1 points to structure 1 where the type is “digit string” and the data pointer points to “31415926….”
How big is the program?
I count two pointers that could be just one byte only each…
You could have one heck of a lot of output certainly way more than you might think from just two bytes
In fact you get the same using “byte code” interpretets where one byte is the equivalent of “PRINT ” and the second “$STR1”.
So the reality is any string of digits can be represented by a single byte code plus the upto infinite output string followed by a terminator. Or with a diferent byte code a filename string where the file contains an upto infinite output string…
Which is the valid program size you are using?
It’s why you have to compare “Apples wirh Apples” which mostly those presenting their arguments or proofs do not…
@ Winter,
Which brings me to your,
You quoted Cantor as proving there are more reals than natural numbers.
In the mathmatical sense of Cantor’s argument, there is an infinity of reals between each natural number.
However in the infinite string of digits sense all those reals collapse down to the same as the natural numbers, as you would expect.
Winter • June 19, 2022 9:58 AM
@SLF, Clive
I am talking about deniability under a state of betrayal and you are talking about some form of “Key Managment(KeyMan) that involves “Key Generation”(KeyGen) and “Key Distribution”(KeyDist).
I think the nicest feature of an OTP is that I can send an encrypted message saying, e.g., Trump plans to get Pence Murdered on January 6, and then can generate an OTP that decodes the message to, I plan a surprise party for Jane on January 5.
Whenever I am forced to show the secret key, I can give them a valid key that decodes the crypttext to any clear text I want. And if Bob claims otherwise, he has to prove that he did not make his key up.
Winter • June 19, 2022 10:59 AM
@Clive
“What is the difference as far as printing is concerned, of an integer of infinite length and a real of infinite length?”
If you can prove Cantor wrong, prepare yourself for the Fields medal.
Winter • June 19, 2022 11:57 AM
@Clive
However in the infinite string of digits sense all those reals collapse down to the same as the natural numbers, as you would expect.
Cantor also showed that talking about infinite set is very tricky.
I feel not qualified to explain the differences, others have done a good job here. But a fundamental mistake in your reasoning is that the integers are part of an infinite set of finite sequences of digits. The reals are a “larger” infinite set of infinite sequences of digits.
lurker • June 19, 2022 3:04 PM
That must be why those who don’t understand call it “Cantor dust”.
SpaceLifeForm • June 19, 2022 3:42 PM
@ lurker, Winter, Clive
The size of infinities is infinite
Aleph Naught went to the Hilbert Grand Hotel
She was able to get a room.
hxtps://en.m.wikipedia.org/wiki/Hilbert%27s_paradox_of_the_Grand_Hotel
But, the bar was very crowded.
hxtps://nitter.net/cwsshotboard
Winter • June 19, 2022 4:01 PM
@SLF
The size of infinities is infinite
Set theory is fascinating, and I marveled at Cantor’s theory when I first saw it. However, infinites have limited practical value outside mathematics.
The case of finite sequences is much more useful. And the link between computational complexity and probability theory is of real practical value in this.
Clive Robinson • June 19, 2022 5:01 PM
@ Winter,
If you can prove Cantor wrong, prepare yourself for the Fields medal.
I would have no wish to be “so blessed” or for that matter inflict such “a blessing” on others. Living in a floodlit goldfish bowl is not something anyone should have to suffer.
But my intent was to point out that there is a very real difference between the theory of numbers and the practical reality of expressing numbers in a tangible form no matter how long.
The point being that a programme deals with the practical “tangible” realities of life, where as number theory all to often can not be expressed in a tangible way no matter what.
It’s why I talk about comparing “Apples with Apples” and why if you can not do that, then your proof no matter how apparently elegant is not anything of the sort. A mistake that is oft made and I oft wish I did not come to my attention.
Cantor’s “diagonal argument” is not in the slightest touched by this because “It works with apples through out”.
But getting to the point of not just knowing this but understanding it in a way that enables further insight is a step few make, or apparently can make. In fact one of the issues we still do not understand from Cantor’s time was “madnesss in mathmaticians”, their early demise, and why there was a strong corelation with certain heriditary groups from the areas of Austria and Hungary of the time.
SpaceLifeForm • June 19, 2022 5:36 PM
@ Clive, Winter
“What is the difference as far as printing is concerned, of an integer of infinite length and a real of infinite length?”
Well, if it is an Integer, then it can not be infinite, because, by definition, it is fixed. It is not an irrational number that has infinite digits.
Cantor was not in the printing business.
He missed his calling.
Fine print in the license: You are only allowed to print irrational numbers on this printer.
Clippy; It looks like your paper and ink are getting low. Confirm reorder?
Clive Robinson • June 19, 2022 5:54 PM
@ SpaceLifeForm, Winter,
The size of infinities is infinite
Actually they get infinately small…
Ask yourself a question,
“How far appart are two points in a space?”
We know that an ordinary measure can be done by taking an arbitary integer N and scalling it to fit.
We also know that between each integer of N is an infinate number of reals. So N x Infinities, with the distance between them being 1/(N x Infinities). So far so easy you would think…
But what lies between each real? And what is there spacing?
So what is the actual space between the reals?
But more importantly how do you tell one real from another real?
Time to find good room service 😉
Winter • June 20, 2022 12:10 AM
@Clive
Actually they get infinately small…
Cantor is famous also because of the massive backslash of his theory. Infinites have driven people into religion and madness (whatever the difference between the two).
It is not productive to speculate about infinites just as it is not productive to speculate on quantum theory.
SpaceLifeForm • June 20, 2022 3:12 AM
@ Winter
There is a fine line between Genius and Madness.
According to Quantum Religion, this fine line is not Infinitely Small, but has a limit called Planck Length.
Subscribe to comments on this entry
Leave a comment
Sidebar photo of Bruce Schneier by Joe MacInnis.
willmore • June 9, 2022 2:53 PM
Isn’t it Thursday? Maybe you’re traveling and it’s Friday where you are.