Schneier on Security

Clive Robinson • January 10, 2016 7:30 AM

@ Curious,

With regards David Chaum’s proposal, jokes asside, and the obvious issue that “the nine” would become vulnerable to all sorts of attack, the protocol appears to be susceptible to side channel / meta data attacks.

So my advice with regards it is currently “get yourself a very very long spoon, or not accept the diner date in Hades”.

The important thing is no mater what backdoor you put in for Law Enforcment there are going to be non crypto ways around it. In the case of David Chaum’s system it only takes a tiny change in the law and refusal by one of the nine would mean “a big chunk of time in prison” if they are lucky, considerably worse if they are not.

ter all the US Gov amongst others has shown no reluctance to either torture or have tortured people for no justifiable reason as well as detaining innocent people with out charge, as they have become a political embarrassment, then there is of course the arguments about standing that judges go along with for a quiet life…

As Bruce has noted Backdoors are a bad idea, I’d add that they are not just a bad idea but a very bad and very dangerous idea, from which no good will come.

Clive Robinson • January 10, 2016 2:31 PM

@ Nick P, Wael,

Beware the ides of Trumpeting.

Not as widely known as it should be, Japan has a housing problem, their population is aging but their birth rate is incredibly low. Appart from issues with how pensions medicare etc are to be funded there is the problem of “Vacant Property” that has little or no worth, as there are not the people to live in them.

At first sight this is a “Japan Only Problem” but it’s not. It infact effects most Western Countries including the US.

Put simply the average birth rate in those who have lived in the West for more than four generations has diped to or below the expected death rate at birth. Thus it is only those who are living beyond their life expectancy at birth that are holding those populations at close to unity. The problem being that in effect their “economicaly productive life” is now at best one third of their total life expectancy.

Thus for Western Countries the only cause for population stability or expansion is “immigration”. Thus Japan’s issues with property will quickly become the Wests problem with property. Then of course there is the cost of medicare and pensions to resolve, some say that if there is no jobs/immigration to pay for it there will be significant issues within a decade…

Thus Trumpeting on about keeping imigrants out is in effect “slow suicide” for Western Nations.

Any way an article on Japan’s housing with appropriate info relating to the US,

http://www.citylab.com/housing/2015/12/what-the-us-needs-to-know-about-japans-vacant-property-crisis/422349/

Clive Robinson • January 10, 2016 5:52 PM

@ Me,

It is infested with JTRIG types and anything of concern to us is flagged out-of-existence, buried and left for dead.

The UK GCHQ Joint Threat Research Intelligence Group (JTRIG) has existed for a long time, they were certainly knocking about back in the early 1980’s “doing Maggie Thatcher’s biding” as were some other organisations (hunt out the history of the Miners Strike and some of the farcical things that went on).

It has been said that JTRIG were successful at manipulating the US Gov over the Falklands war (it’s no secret that the US promised the Argentineans the Falklands in return for turning a blind eye to what the US were upto in Chile and back and forth across the border).

The problem for the US was that Maggie was in favour of what was going on in Chile rather than Argentina, and rather disdainful of what the US were upto. It was for some time testing “The Special Relationship” to almost breaking point (again it is no secret Maggie Thatcher despised some of the US political machinations which is why she got rid of the War Debt that the US had used as a blackmail weapon against UK technical and scientific interests).

Various historians say it was Maggie’s “Charm Offensive with Ronnie” that won over America, however I doubt that, that alone would have succeeded. So there may be some truth to the JTRIG Falklands War stories (but as always take with a “Lotts Wife” sized grain of salt).

The thing is ever since Sefton Delmer[1] put his foot in it and moved over to do “Black Propaganda” in WWII there have been various groups doing this sort of thing. They obviously, as such groups do, make outrageous claims about their supposed successes, but you don’t hear about their failures etc, nor do you get to see how they calculate the magnitude of their supposed successes.

Which is why I suspect that if audited correctly they would prove to be an expensive but ineffectual organisation.

As for where they appear, I would not assume that this site is unvisited by them or their counterparts, in fact I actually think it’s quite likely they do.

The art of Black Propaganda is to “tell the truth” but from a particular perspective, then put a further slant on it with unverifiable opinion etc. So you might have a comment consisting of verifiable facts that are both easy and hard to verify and an opinion that is neither supportable by the surounding facts or verifiable in it’s assertions.

If you look at some of the comments people make on various threads you can see this behaviour, sometimes well sometimes not. Which leaves the question of if it is organised or just individual agenders, and further if organised if it is State or party political, NGO, Commercial, etc. Some of it is so. obvious that it gets called out by other posters and even this sites owner and moderator.

[1] https://en.m.wikipedia.org/wiki/Sefton_Delmer

Clive Robinson • January 11, 2016 11:39 AM

@ Curious,

I wish someone could explain to me why a 32bit nonce is so much more insecure than a 20bit nonce.

As the article says it’s bytes and it’s output from the Dual EC DRNG. Unfortunatly it in effect gets “passed through” to the network thus giving the attacker (NSA or whoever) an extra 12bytes or 96bits of the EC RNG output. As a very very crude aproximation the attack time halves with every bit so you reduce the work factor to crack the EC RNG from infeasable in any realistic time frame to doable in a few seconds.

I hope that helps.

Clive Robinson • January 12, 2016 4:56 AM

@ Mark H,

This really depressing Computerworld article explains in some detail how bad the security practices are among vendors of the major anti-virus packages.

If only it were just “major anti-virus” vendors.

As I’ve pointed out in the past “code signing” says nothing about the quality of the code or of those that designed / developed / tested it. In fact it can be shown that the major benifit of it is actually DRM on non open platforms, of which games consoles have been the most obvious for the longest time, followed by the “walled gardens” of smart devices like phones and pads as pushed by Apple, Google and Micro$haft.

It is a battle we are losing and it will be slamed shut within a few years judging by what is in the TPP drafts.

With the advent of near compulsory 24H “fit-bit” wearing by employees of C Level execs looking for cheap medical insurance, I can see a time when the US Government gets mandatory access to the data.

Then as I’ve indicated in the past as part of faux “tax reduction” measures people will be very much increasingly “fined” in new and inventive ways by the Governmentoff of the back of such insurance systems. In the UK we have seen proposels for reduced “road fund” based on having your car “wired in”, and thus as with tacho’s speeding and other road offenses will be easier to “find and fine”.

Thus with the likes of new and improved “Well-bits”(TM) a $100 fine and mandatory costs for the misdemeanor of, breaking wind, coughing, sneezing or even breathing to deeply etc in public places will not just occure to the money grubbers but be easy to implement by dictat.

It ceases to be mildly amusing and become the stuff of nightmares when you think about Smart Meters, IoT and Medical Implants…

As for privacy the only solution would ba a three step process, single sheets of paper on glass surfaces and soft pencils to do first stage encryption, then an “energy gapped” old PC to get it into “electronic form” and second stage encryption / normalisation / obsfication / stenography then transfer via some kind of one way process to the “Snoopernet”, all whilst practicing solid OpSec to avoid IoT in your domestic appliances evesdropping and shoulder surfing 24×7.

That glowing red eye of HAL in 2001 lip reading Dave is where IoT is heading, unless we find ways to stop the Police/military Industrial Complex and their machianations. After all why have an expensive RoboCop when you can have a built in “White Good Spy”.

Time to stock up on “old tech” whilst it’s still yours to own and control. The fashion statment of “So last Century” might just apply to the new breed of “Wise Guys”, not just those with a retro vibe.

Clive Robinson • January 12, 2016 9:05 AM

More on Tor… and CMU

Some of you may know that back in December, Tor hired Shari Steele, who was for 15 years previously the EFF executive director. Shari is Tor employee number ten and has been interviewd by Arstechnica,

http://arstechnica.com/security/2016/01/going-forward-the-tor-project-wants-to-be-less-reliant-on-us-govt-funding/

The interview covers several things including Tor’s desire to diversify it’s funding, so that the development they can do is not constrained as much by Project Funding from the US Gov which effects what they can and can not do. It also covers the deliberate avoidance by CMU over it’s FBI sponsored attack on Tor, and why CMU behaving in such a way has implications way way outside of Tor and right into the heart of day to day Comouter Security, especially on things like the rapidly declining trust in CERT.

On a point that’s not mentioned in the article the FBI-CMU-Tor debacle also raises the question of “ethics”. As some know all research carried out in Western Universities should be subject to scrutiny by an independant ethics committee (this is a legal requirment in some places). It is now abundently clear that CMU failed in this vital respect as at the very least it kept PII records that could be subpoenaed, and this has without doubt brought down 2nd and 4th amendment issues onto innocent heads, some of whom were carrying out legaly privileged activities.

Thus it should now be beholdent on all ethics committees especially those in CMU to ensure that all PII and Meta Data that can give rise to PII should be suitably anonymized at the point of instrumentation (sniffing) prior to it being put on any communications link, recorded, processed or collated in any way. And that if data and meta data that would give rise to PII can not be sufficiently anonymized then the research should not go ahead (this is the norm for the likes of human medical trials etc). Further all funding returned and any work carried out to that point should, if it would enable others to follow in the project foot steps sufficiently to replicate the PII revealing activitied should be made unavailable academically or legaly and these requirments made absolutly clear in any funding proposals.

Whilst it might sound draconian it is still more than possible to do research into information security that would reveal and allow PII threats to be identified and fixed by a variety of fairly well established techniques, not to disimilar from those used in malware research etc. That is there is very few cases where “live systems” need be instrumented in a way that allows PII recovery, and any such proposals should be deeply scrutinized by independent experts in advance of such work to ensure the ethical constraints are suitably built in.

Clive Robinson • January 12, 2016 5:25 PM

Microsoft Windows Remote Desktop Protocol Security Bypass Vulnerability

On the day Bruce posts about Apple -v- Microsoft vulnerabilities last year…

Microsoft pop this one out,

https://technet.microsoft.com/library/security/MS16-007

Which is a fun start to the year…

It contains several major vulnerabilities but the one bypassing login security on the RDP for accounts without set passwords in Win10 is a real doozy, as it looks fairly simple to exploit.

Clive Robinson • January 12, 2016 5:37 PM

Accenture survey finds 47% of people find “security concerns” in the top three reasons not to get IoT devices

http://venturebeat.com/2016/01/04/mobile-device-sales-slow-customers-grow-wary-of-security-holes-in-connected-devices-survey-says/

Looks like the message is slowly getting through to consumers afte half a decade or so of warnings.

Clive Robinson • January 12, 2016 7:41 PM

@ Nick P, Wael,

You two up for a side bet on “bragging points”?

As you probably know Pres Obama is giving his last “State of the Union” address in a short while.

Obviously he is going to mention terrorism and IS, but will he also mention technology and encryption?

And if he does which way he’s going to push it?

Now a little bird has mentioned in my direction that the word is out that when Obama met with Silicon Valley tech last friday Comey pulled a switcheroo on the agender and banged on about encryption not how to deal with IS.

Unless the little bird who gave me the heads up is from the same nest as one Bruce has been listening to then other people are hearing a similar story that Comey is about to get some legislation that is leaning his way.

Do you think Obama is going to mention “a moving forward” on this?

Clive Robinson • January 13, 2016 2:52 AM

@ tyr,

How clothing makes
them come together sounds pretty questionable.

There are oh so many ways… To answer that, but I think I might get banned if I did… Although Internet traffic type figures suggest there might be many “amateur researchers” looking at the issue but few of them get into physical research or publishing 😉

@ Nick P,

I’ve not got any science papers, but I do remember a talk on evolution that looked at diferences in “native populations” who lived in differing environments. And there were plenty of interesting examples like Nepalese and their ability to exist at high altitude low oxygen environments due to differing blood. Chilean’s and their square thigh bones giving better muscle anchoring. And native Australians who had slightly different eyes that could see a lot further, but also dropped their core body temprature by half a degree C when sleeping, thus could sleep outside more efficiently.

But the one that “blew my socks off” was in arctic canines where their leg blood vessels were different in that the veins and arteries were not just closer together but actually entwined, thus making an efficient heat exchanger.

So there are examples of different thermo regulation stratagies developing in issolated groups in response to environmental conditions they live in.

From other research there also appears to be other evolutionary effects in testicles in primates. The more monogamous the species the smaller the testicles are, suggesting that “flushing” is an effective evolutionary response.

So yes changes in social behaviour happening orders of magnitude faster than environmental movment might account for a lowering in the birth rate.

However, it’s also known that in other societies which switch from agrarian low health care to industrial high health care, the fertility rate drops from an average of five children per couple to two or less. And that the age of child bearing changes from teens/twenties to thirties/fourties when fertility is naturally lower.

But even intelligence appears to effect fertility rates, there is an inverse relationship with standard IQ measures and fertility. The higher your IQ the less likely you are to have children and back in the mid 90’s there was a study that showed above 130 the fertility rate is around 1.3 children per couple versus 2.7 at 110 rising to 6 below 90. Giving rise to the argument that “Intelligence was anti-Darwinian”. Interestingly another study found that there was a socio economic status effect in the higher IQ fertility rates. The wealthier you are the more children you have and that these tend to be men having 4 or more children with two or more different women (the so called “breeders and squeezers” of “trophy wife” culture).

Clive Robinson • January 13, 2016 4:31 AM

@ Wael,

I’m aware of the German saying and three of it’s “supposed” meanings (though just trying to pronounce it should give a base translation for those that speak English but not German).

The variety of meanings is typical of German sayings there is one that literally translates as “Warm Shower taker” which has a variety of meanings one of which would be the same as if you left Gut (good) off of the one you mention, or at it’s mildest some one is a wimp, but could also cover the Japanese hot springs users…

However as far as expletives for expressing pain or frustration, in that neck sinue stretching teeth grinding way, Norway has a few that are hard to beat. They are actually quite mild in meaning kind of “oh drat” level but you can realy put your all into them and they sound realy emotive to someone who does not speak the language. The one slightly rude word that makes me smile is “Rumpentute” which most can guess refers to a horn like toot from the bottom. It’s the sort of word you would expect a six year old to say and giggle over, but it realy resonates when arising volcanicly from the depths of the lungs of a man for whom wrestling grizzly bears would appear to be mild excercise.

Clive Robinson • January 15, 2016 3:48 PM

London’s Met Police spying with fake cell sites

https://news.vice.com/article/vice-news-investigation-finds-signs-of-secret-phone-surveillance-across-london

Clive Robinson • January 16, 2016 2:30 AM

As somebody else has mentioned Cory Doctorow, I guess I should mention this article,

http://boingboing.net/2016/01/12/keep-your-scythe-the-real-gre.html

Whilst it is a kick out at the “lefty forward to the past through rose tinted glasses” idiots, it also highlights another more serious issue which is the elites “Politics of status”. In the process it also gives a simple explanation of why the “Free Market” mantra is wrong, and how economists deliberately blinker themselves to promote the view their sponsors want to be highlighted.

At the very least it is an amusing read.

Clive Robinson • January 16, 2016 5:58 PM

For those who taje notes on US Police using excesive force for effectivly no reason,

http://sfist.com/2016/01/14/black_bicyclist_says_he_was_stopped.php

Clive Robinson • January 17, 2016 5:54 AM

Hotdesking gets junk imaging

If this were the first of April I would think it a badly done April Fools Joke…

Sadly it appears not,

http://www.theregister.co.uk/2016/01/15/put_your_private_parts_on_display_if_you_want_to_earn_a_living/

Put simply a battery operated imaging device [1] is put under the hot desk to view your groin to see if you are at the desk. It sends information off wirelessly to a central system where your groin image is logged in the cloud for cross refrencing by Human Resources or others with cloud access…

So if you are one of those “side saddle” or “stand up” workers, you will be regarded as “not playing” so your work record and ultimately pay could be adversly effected.

[1] http://www.occupeye.com/how-it-works/

Clive Robinson • January 18, 2016 12:38 AM

@ Sue Dough Nym,

How Encryption Technology Can Be Traced Back to the Kama Sutra

David Kahn wrote about this in his 1960’s book “The Codebreakers”.

From memory –if it’s not failing me– one cipher was based on phonetic variations in the words, the second cipher was a simple reciprocal substitution which was effectivly a simple reflector where you wrote down one line above another for the Key with half of the alphabet characters in the top line and the remaining half in the bottom and you simply swaped the verticle pair in use. Thus the same key was used for encryption or decryption without change in proceadure.

Like the later Enigma machine and it’s wired reflector it solved one –human– problem by creating a worse –technical– problem. However the Kamasutra was written befor the Arab Scholar and mathmatician Alikindy –from where we get the word algebra– so how to break such a cipher was possibly not known then, but the same was not true for the Enigma invented shortly prior to WWI…

Clive Robinson • January 20, 2016 6:54 AM

@ Nick P,

Dr Royce is right about many things, I did not find out about him untill after I’d got the better part of a quater of a century of engineering under my belt.

In that time I found out for my self about programing and it’s relationship to art (Donald Knuth has some choice things to say on that as well).

My viewpoint is that whilst there are parts of programing that are art and science, engineering was most definitely not part of it (I was an engineer that took up programing as a necessary evil 😉

On studying a bit of history I realised that most programmers are actually very much the equivalent of Victorian “artisans” that would later with the help of the scientific method become engineers. The only reason the Smiths and wheelwrights became engineers was due to the carnage of boiler explosions and the outrage that gave rise to legislation and thus licensing, with formal rigours “technical training” not just informal and patchy “mentoring apprentices”.

It’s why I have found that those who have trained through the rigour of other engineering and science qualifications tend to make better programmers when it comes to interfacing with the real world not just users. They learned as past toolmakers did how to build tools to carry out other activities. Thus they hit the ground with running feet not theory on the psychology of user interfaces (yes I know people will take exception but hopefully they will think on it first).

It’s taken years for formal methods and engineering design practices to creep in alongside “artisanal patterns” etc and for others to realise the benifits, and it will still be “pushing in” long after I’m gone unless Governments make legislation that requires the formal methodology.

As for documentation yeh… It’s vital in oh so many ways. As I’ve mentioned befor I comment code very heavily. So much so that it’s been pointed out a few times that I document twice and write code once. And yes it’s true, you can read the way the program functions at the interface level in the sub header commentd or on the line by line level adjacent to the code, as well as read the code it’s self. It’s a habit I got into writing at the assembler level and below, because it sanity checks things and makes fault finding much less difficult.

Then there are “History Files” few people make them even when required to do so. Which is a shame because they are a gold mine of information both you and others can refer back to. Oh and on the personal not project level, both scientists and engineers keep log books / diaries of what they have done and why and experimental results. They get to do this because they get told “If it’s not written down it never happened”, but quickly find that it becomes a personal guide and refrence book. When you look around you will find productive programers keep diaries / log books as well as history files.

Also those fron non CompSci degrees tend to develop styles that Dr Royce wrote about they kind of found out the hard way that it works…

Clive Robinson • January 22, 2016 5:47 PM

@ Wael,

Your eclectic tastes cover such things as the “Fermi paradox” sometimes called “WTF are the Aliens?” question…

Well you are probably aware of what some say about “life is to short” in the Goldilocks zone, unless you have a moon, and giant outer planets.

Well down in Auz at Parks they have another similar but different theory,

http://astronomy.com/news/2016/01/the-aliens-are-silent-because-they-are-extinct

Personaly I’m happy with the “hundred years of broadcast radio” theory. Which basicaly says that it takes a civilisation about a century to go through discovering, then exploiting radio first with high power broadcast systems with simple modulation systems that might be heard outside their solar system, to using complex modulation on low power personal systems that can not be detected at one astronomical unit from their planet let alone at the outer edges of their solar system or beyond their suns heliosphere.

A hundred years is but the blink of an eye when compared with how long the dinosaurs walked this Earth.

Clive Robinson • January 24, 2016 4:00 AM

@ Nick P,

<

blockquote>Most of their stuff was discredited. Others postulated a lot but nobody has evidence.

<

blockquote>

Not exactly discredited nor is it the case that there is no evidence. Both are signs of a very new field of research moving from theoretical beginings.

There is growing evidence that biological mechanisms use mechanisms we can not explain except by quantum effects. So we now have the field of Quantum Biology, the problem is not that the processes may be quantum, but finding the process mechanisms [1] and finding out how they work which is the job of the experimental scientists.

And as you would expect in a new field of endevor there is a lot of whittling down of false starts / leads to do. You would not expect an original Peeler (Policeman) to be able to do what modern Crime Scenes Investigator teams can do, because you need to build up a body of knowledge, practices and tools.

Quantum Biology is becoming a recognised part of life, for instance plant photosynthesis can apparently reach 99% efficiency in energy conversion across a broad spectrum that can not be done in any other way based on our understandin of just a few years ago. And we are now begining to get to grips with the mechanisms.

Further there is the “mint / orange” issue with smell/taste, that likewise appears to be resolvable if the “nose” worked at the quantum level, with secondary arguments about other creatures being able to detect single molecules of odor compounds again requiring an efficiency that could not be obtained in simple chemical ways.

The main stumbling block appears to be “time” traditional understanding of biological mechanisms is that they are very very slow, and quantum systems very very fast. With the difference in some cases being in the order of 10^8, thus we are looking for efficient cascade effects. The problem there being that most macro biological systems are only moderately efficient in cascade –think light power to hay, to horse, to cart, to mechanical power– thus we are likely to be looking at the molecular scale of things.

One of my local Universities (in guilford surrey) –more world famous for space payloads– held a workshop on Quantum Biology a short while ago and have put some of the presentations up,

http://www.ias.surrey.ac.uk/workshops/quantumbiology/report.php/

You might recognize one or two of the faces of the presenters.

[1] @Wael, that efficient rotary engine that fascinates you in some cases uses a proton pump mechanism which pushes it into the growing Quantum Biology coverage.

Clive Robinson • January 24, 2016 6:44 AM

@ Wael,

How would you suggest we model ‘natural selection’?

The answer is simple but brings up a new level of complexity that is far from simple (just like weather forcasting).

You model the dynamic environment and the pressures it creates for your chosen subjects.

The normal way to do such a model is to have a surface that you grid up you then make each cell in the grid dependent not only on the contents of the cell but it’s neighbouring cells as well.

The problem is deciding in what order to update each cell, because any change in one cell is reflected back to it via it’s neighbouring cells as well as the cells neighbouring those. That is every change ripples out as well as reflecting back.

If things are linear in behaviour then you can make assumptions based on what happened in the previous itteration and that can minimise the effects of ripples and reflections caused just by the selection order. But first you have to analyze the algorithm to see how it behaves.

You thus run the model a hundred or so times with small adjustments in the starting conditions and try to work out what is linear and what is not and how sensitive the outcome is to the changes. You then feed that knowledge back into your model and run it a hundred or so times again and try to isolate the non linear effects to see if they are chaotic or not. Where there is chaotic behaviour try and map out the cusps or tipping points.

That should enable you to get a feel of your model to see if it is realistic or not and what the dynamic behaviour is. Thus any computational short cuts you can take.

The most likely thing to come of it will be an interesting game for others to play. That appears to be the fate of most models since the original “game of life”.

As for the cell algorithms you can use an energy transfer function based on the second law of thermodynamics. In effect you are looking for algorithms that are very local that take one of many coherent energy sources and as efficiently as possible builds coherent self replicating structures. Obviously as the environment changes then the most usefull energy source changes. Which means there is a trade off between becoming two focused on one energy source and becoming extinct as and when it does or being over general and thus not sufficiently efficient to survive. As an example it’s advantageous to store energy but that takes volume and mass that effect your mobility thus making you an energy source to others. Thus you have to also decide on the trade off between survival of individuals and the species by balancing the replication rate with the mobility rate derived from energy storage to predation rate. Apex preditors effectivly don’t get predated so favour the individual over the species and thus have low replication rates. Those towards the bottom of the pile have very high predation rates thus have high replication rates. Plant’s and some animals try to cover both by also having self regeneration properties. Apples are of interest here they have rapid diversity on replication, thus when a breed of apple is found that has marketability growers replicate it by cutting and grafting and using the plants regenerative abilities to make new plants, with no diversity.

Now you can see why there is a fun game like element to such models.

Clive Robinson • January 24, 2016 2:11 PM

@ Wael,

That may work for a first iteration ‘model’. Assume everything is linear, ignore interdependencies and mutual interactions,…

Err not quite the first iteration is is to find out if there are any “hidden” or unexpected effects.

Still, I don’t think we took everything into consideration:

Actually we should taken very little into consideration with the first few iterations. The idea is to find the sensitivities and ensure that the way order we proccess the cells etc does not impact on the model to the point it starts to effect the results.

Problem is: trial and error implies intelligence,

No it does not, not in the slightest. Take the ability to grow, this is dependent on getting sufficient energy and not getting predated by other creatures, no intelligence required there. Likewise the ability to grow improves mating privalages, again no intelligence required there. After all how much intelligence does a microbe or weed need to be successful not just as an individual but as a species?

All that is required is the ability to feed and breed, with occasional mutations in individuals effecting feed / grow / breed abilities that get passed on down the generations.

Can we apply the same principles to security and expect a “Secure. System” to evolve out of that process?

Depends on what you mean by “Secure System” thus how you set the equivalent feed / grow / breed goals. Which might be your hardest problem to solve.

Clive Robinson • January 24, 2016 3:36 PM

@ Wael, Nick P,

I know you guys just cannot get enough high quality reading material for free. So to stop you wasting money or twiddling your thumbs you might want to read this free eBook on Neural Network Design,

http://hagan.okstate.edu/NNDesign.pdf

It’s just over a thousand eBook pages so should keep you quiet for a few weekends 0:)

Clive Robinson • January 25, 2016 2:26 AM

@ Nick P,

What jumped out at you?

The issue of “inexplicable results” or if you prefer “results pulled from a magicians hat”.

As you note,

How the hell would we even start understanding (a) why they worked at all or (b) why they weren’t working in specific cases? That’s still an open question and worth avoiding them for high assurance applications outside maybe a first stage w/ checks following.

The problem is not just “high assurance” it’s anything where there might be a legal liability.

As has been pointed out in the past one of the important reasons we don’t have driverless cars is not that they are technicaly difficult, but the legal implications of a “directing mind” based on a thousand years of lawmaking.

Humans appear to have an inbuilt fear of the unknown, even when we in fact know it’s safe (think darkness in a bedroom and the creak of heat loss in the building structure causing people to assume monsters / wild animals etc). Many have indicated it’s the primative monkey brain warning us we are just about to get eaten if we don’t “flee for the tree” and others have tried to explain it as “the intervention of God”. The thing is though we don’t understand it and it causes a hightened emotional response in nearly every one, especially when an immediate explanation of the noise is not apparent.

Such emotional responses often cause other behaviour such that people will point fingers and scream “burn the witch”. Or in more modern parlance “sue for damages”. The only defence to this is usually to show negligence by another or a highly improbable event that gets called “an act of God”…

Worse humans tend to “over trust” other humans. Accident statistics say that the major cause of accidents is not mechanical failure or design flaws but human operators (not being sufficiently trained, or over estimating their abilities). Faced with the facts it’s amazing we actually get into vehicals in the first place…

Another human failing is to put desirable human characteristics onto animals and machines… Why we do it is way beyond rational logic. Thus you genuinely do get people who have vehicles in a poor state of repair beliving that their “good little car, would never hurt them”.

Further is the unscientific arguing from effect to cause. When an improbable event happens people claim it as skill or lack of forsight depending on good or bad result. That is you “get a hole in one” you are a skillful golfer, you go for a walk with a girl on a sunny day the heavens open up she gets soaked it’s your lack of forsight because “you talked her into it” or “you didn’t bring her coat / umbrella” etc etc.

In short people do not behave rationaly even in normal circumstances, and will seek to blaim others when their irrational beliefs cause them or those close to them harm, and if they can not pin it on somebody then a deity will do, just as long as they don’t have to take responsability.

Which gives rise to a problem, you have a device that people start to trust and it causes then harm, you know that there will be an accounting. You will have to argue that you are not just “not at fault” but also “not culpable” in any way… Which means even if you can demonstrate the operator was behaving in a way that was unwise, you have to show that you were in no way responsible for the unfortunate result they suffered. That is they expect the impossible for you to have 20-20 forsight.

Do you realy want to stand up and say “We do not know how this works, but it tested OK”, you just know in advance that some shyster is going to claim it’s your fault for not testing adequately or in this particular set of circumstances etc etc etc, it’s why we call it a “loose loose situation”.

But it also has another aspect, clearly the result of the mathmatical process is a usefull object, rather more complicated than a watch. How do you get around the “watchmaker argument” of “inteligent design”?

If you can not demonstrate “the how ” of a usefull object, then in most peoples eyes you are not the designer. Therefore someone else inteligent “must have” designed it. But what if nobody can be shown to know “how” then a few people irrespective of what evidence you produce about the process will still look for inteligence not probability and thus see “the hand of God” or other “divine creator” such as “the work of the Devil”.

Either way they are insisting in a belief of “intelligent design” not “directed probability”. The problem with such reasoning is “who created the watchmaker”, it’s a “lesser flea” problem[1] which can have unfortunate results in those who chase after it[2].

Which has in turn another issue which is “not being able to know”. Some people will not accept there are limits on what they can or should know. It’s seen by some that they are somehow lessened by this that it robs them of destiny or power etc etc.

I’m known for saying that “There is no such thing as an accident, only lack of timely knowledge”. People hear the first part and get upset, rather than think on what the second part means. We are seeing for real this issue with “Big Data” the assumption is “if we collect it all we know it all” therefore acts of terrorism can be stopped. They forget a few things,

1, Information is not knowledge.
2, You have to be able to turn information into knowledge.
3, It takes time to turn information into knowledge.
4, You have to be able to use the knowledge.

The last one on the list is the big problem. Prior to 9/11 it was known that a flight school had reported odd behaviour in some of the trainees. Prio to 9/11 the knowledge did not have context and thus made no real sense. However post 9/11 it was easy for “armchair quaterbacks” to go from effect back to cause and everybody pointed the finger at “Intelligence failures”.

It appears not to matter how often you point this out for various reasons people don’t want to take on board the idea that “not being able to know” or “imperfect future knowledge” is an intrinsic fact of life. Thus even in a “pre ordained universe” life is probabilistic and does not need “the hand of God” to throw the dice.

[1] From “Dogs have upon their backs fleas to bite them, upon these fleas are lesser fleas, and so ad infinitum”.

[2] Serious contemplation of the infinite has given rise to the very real problems in the past, of mathmaticians becoming seriously unbalanced if not mad and dying much earlier than they should have done.

Clive Robinson • January 25, 2016 7:30 PM

@ Wael,

So you are in effect using a sive / screen model as a fitness gate, and an initial random generator.

What I don’t is how you intend to get further changes?

That is the usual way is to either,

1, Take two objects that have passed through the first screen and randomly select a combination of their parts.

2, Take a single object that has passed through the first screen and make a copy but randomly change one or more of it’s parts.

You then take the whole lot and run through the second sieve / screen and repeate the excercise.

Importantly the second sieve / screen is it’s self a mutation of the first sieve / screen but moving in a chosen direction to represent environmental change.