Schneier on Security

Clive Robinson • August 5, 2021 3:51 PM

@ Truth.is.stranger.than.fiction,

I hope you are feeling better and are home.

Yes to the latter, and whilst I feel better physically, mentally there is quite a bit to think on… And supprisingly it has lessons for security people in it as well…

Basically I’ve a blood clot 0.8 inches long and 0.7inches and it’s in my right atrium. The output from my Left ventrical in maybe 20% of what it should be, and officially I now have “heart failure” to add to the list… Oh and several blood clots in the lungs again, and a compleatly new one for me “gaul stones”…

So lesson 1, “Life can always suprise you”.

For those that do not know the “cure” such as it is for blood clots, is what some call anti-coagulants and others call blood thiners. Me being more of a shovel is a shovel type person call it what it realy is in this case “Rat Poison” (much to the surprise of many in the medical proffession). You can look it up as it’s called warfarin and untill recently it was killing rats by the bucket load[1] but due to security lesson number 2 it’s not…

Well back to the blood clots, there are four basic things that can happen with the blood clots,

1, They get worse or don’t reduce.
2, They reduce slowly and get reabsorbed by the body.
3, They break up and go every where you don’t want the bits to go.
4. They detach and become a major obstruction at say the heart valves.

You don’t want 3 or 4 as you’ve less survival options than the rat. What you want is option 2 which option 1 can be turned into by carefull dose control.

The thing is staying in option 2 effectively is a bit like a jugling act whilst tight rope walking for many people. It should not be but it is and I reckon it’s due to “scary maths teachers” and the evils of “infinitesimals” as Sir Isaac used to call them.

The reason is what is given the grand sounding name of “Half life calculations” which sounds like it belongs in those “physics labs” the military make their bombs in. In teality it’s a very very natural and almost the most basic of real world calculations.

Rat poison like most poisons has a unique effective dose for every individual and it changes for various reasons. It also like virtually all chemical and biological reactions has a “half life” the calculation of which for some reason causes some people to get twitchy, when realy it’s just a “percentage of the remainder” calculation over a fixed time period which realy is not hard to work out. (90% of 1 followed by 90% of 0.9 followed by 90% of 0.81, 0.729 etc). When you plot the curve out for long enough you will find that at a fixed interval (of time) you will have half of what you had at the begining of the time interval and this goes on in theory for ever. Hence the term “half life” for the (life) time it takes to lose half the quantity.

Third security lesson, nature hates linear more than it hates vacuums, so exponential curves are natural straight lines are not, live with it or weep.

But… As with all things in life, nothing is ever simple… The same with rat poison and nearly everything security. Warfarin effects multiple clotting cycles, so although some things are easy to calculate the overall effect is harder… Unless you cheat “constructively”.

This is fourth and perhaps the most important lesson to learn,

It has been said that a cruise missile knows not where it is but where it has been and which direction it went in next. Well the same is true as much for atomic particles as it is for the largest of ships and everything in between including tracking bullets…

As most of us know nearly all ships have the nack of ending up where their owners want them to be without the deck officer using much more than pencil and paper and often just addition or subtraction.

The trick is to estimate where you will be at a certain point in time in the future. Then at that time take a measurment, then work out the difference or “error function” between your estimate and reality. You then feed a percentage of the errpr function back into your next estimate to stear for and so on… Not as difficult as it sounds, in fact I’ve taught ten year olds how to do it, easier than fifteen year olds (I blaim those Maths teachers 😉

The thing is the shorter the time period you use the more those exponential curves actually look like straight lines thus the faster and easier the calculations and the less likely any error will effect the journey.

Thus the real trick is knowing when to make an estimate and when to relax and enjoy as best you can the journey and not getting stressed…

But some journeys should never be made, Dante Alighieri wrote in his “The Divine Comedy” a very notable even for it’s time, graphical description of hell seen through medieval eyes most know as “Dante’s Inferno”. He tried to convey that this journey should not be made back in the last decade of his life (~1309-1321). The Devine Comedy is an allegorical account of his immagined journey through Hell, Purgatory, and Paradise, guided by Virgil and Dante’s idealized love Beatrice.

[1] In recent years what has bern called “Super Rats” have either learned or evolved into surviving. So this is major security lesson number 2 “The environment changes to out evolve any solurion goven time”. Put simply the rats now recognise the early symptoms and go and wrap themselves around a hot water pipe and wait for their body to metabolize and excreate the poison etc.

Clive Robinson • August 5, 2021 9:44 PM

@ Lurker,

What’s so precious about US users’ data that doesn’t apply to anyone else on the planet? The other face of extra-territoriality?

Err “US users” are also “US voters” and “US political campaign donors” anongst other things. They also by right of citizenship have “standing” under US legislation and regulation.

So yeh, as far as US legislators are concerned “US users” are indeed exceptional.

As for “extra-territorial” that’s a more vexatious question and goes back to US taxation laws.

It is still common practice in most countries that they only tax you on what you earn in their jurisdiction, because it generally costs more to chase down extra-territorial income than it actually gains (due to other traditions). The US however considers it’s self to have primacy in such things, thus insists that they receive all the tax US citizens earn and the countries they earn it in should be grateful to not be “Bombed back into the stone ages” and then be told they have to pay for the bombs (remember the Vietnam President being given bill by new incomming US ambassador under Trump?).

Then there are the one sided extradition treaties (remember Julian Assange?).

It’s those little things that get under everyones skin and creates international problems, with US politicos appatently oblivious to the rest of the world.

Clive Robinson • August 6, 2021 6:13 AM

@ SpaceLifeForm,

Hopefully soon, after the Wafarin does it’s job, they can move you to an alternative. Wafarin dosing is very tricky. One day, it is too much, the next, not enough.

I was on warfarin and was managing just fine by myself when I had timely INR readings. It was being moved to the supposadly safer “riveroxiban” alternative that has probably nearly killed me…

In one very nice hospital in East London (who later put a team together and sang their way to the #1 slot in the UK “Top of the Pops”) they had a warfarin clinic system that worked and worked well despite the dreadful slowness of an INR test. The Pharmacist and I used to play a little game on my twice weekly visits. She would tell me what the INR test reading was (given in “prothrombin time” or PT) and I would start doing my mental calculations and she would type it in to their very expensive piece of software that would grind away on the PC. I would say what I thought the adjusted dose should be, and when the computer produced it’s result we’d see it on the screen and compare.

Usually I beat the computer and got the dose right. It was only when there was a big jump that I got it a little wrong, but we finally realised the computer was “ambitious”[2]. So a small correction factor was getting inadvertently added that caused a long tail problem.

I taught the Pharmacist how my method worked and the reasoning behind it and she was a little shocked on the “liberties” it apparently takes with mathmatics. But as I pointed out waves go up and down, so does the tide, and generally ships don’t crash and they get to harbour safely and on time.

Which is the big lesson for security and safety people as well, “Don’t get confused by surface swell, nore draged aside by deep running currents”.

That is after a little thought you will realise that,

“Whilst nature hates linear and will go with exponential in the longterm. In the short term the error function from using “percentages”, is too small in range for nature to care about and rapidly self filters or integrates out[3].”

Thus simple linear maths and “guesses” works fine and dandy and above all is “quick and simple”.

I can give a long winded exlanation as to why starting by noting,

X_n+12 =X_n2 + 2X_n + 1

I won’t this time 😉 because typing in all those HTML tags is almost enough to give a hangover with out the joy of getting one normally…

[1] The INR or “International normalised ratio” test is measured in prothrombin time, or PT. Prothrombin is one of the clotting factors and is a protien produced by the body in that great big chemical factory of your liver. The main problem with this is something called Vitimin-K which in what almost looks like magic stops the effect of warfarin. Vit-K is obtained from a lot of foods, but mainly green leaf vegetables are what you have to be carefull of (which is a real bummer if they are what you like eating, and I do…).

[2] The error function obviously also “opens out” as the size of jump is made. It turns out that the floating point math package the expensive software used had a bug and it got excercised by the particular way they used it… So it was not me that was getting my estimate wrong. NOW it turns out the “bug” made the error work on “the safe side” that is it was wrong in a way that was “ambitious” or “under-damped” and would thus drive the patient faster toward the thereputic value, but then “ring” with a longer “error tail”. My results were slightly over-damped which ment a smoother transition without a “ringing error tail”. You want the former behaviour when you are starting someone on warfarin to get them quickly to the thereputic value, but once stable you generally want the latter as the most likely cause for a jump is a one off ingestion event like a “party” or some such or the pills being taken early or late the day before the test.

[3] Even data scientists can be taught a trick or to about Mother Nature’s Ways. Whilst there is a whole cannon of control theory for those who want to come up with their own method and error functions, the maths grows in complexity like Topsies Cat and it slows to a crawl very very quickly for very very small percentage returns… You can see this all around you but the place you are most likely to see it in action from the hand of man is control loops in “Phase Locked Loops”(PLL). Which you are surrounded by with modern technology, it realy is in all things these days with CPU clocks running many times that of the control Xtal’s.

But the “secret sauce trick” is not to care about accuracy but only about errors and sometimes guesses. You care not a jot where something is actually going, you just make a guess wait a little while and look at the size and direction of the error signal. Then feed that error information back into your next guess and the next error will be smaller and your guess more accurate. Each time through that loop actually also “digitally filters” the result through a low pass filter or integrator which alows you to set any desired “loop filter” characteristics you want. But usually it is “inherently stable”. The advantage of a digital loop filter is you can use diferent guesses to get different characteristics, so a small lookup table keyed by the error output can give you some extrodinary results quickly and easily, whilst the control theory maths will get you there, you will have steam rising from the CPU…

Clive Robinson • August 6, 2021 2:21 PM

@ JonKnowsNothing, SpaceLifeForm, Winter, ALL,

Apple will be using the cover of “Won’t someone think of the children” by actively scanning devices for cop-tagged-files and using other methods scanning for “untagged files of concern”.

So if we take the FUD off, we are left with Apple scanning every file in some way to match a “cop-tag” of some form.

Now you can be sure that some judge will take this cop-tag as,

1, Proof positive because some cop says it is…

2, Not scanning or being looked at under fourth amendment rights because “nobody sees the file”…

Anyone else see where this will go?

How about an embarasing document that says some famous persons child has taken a big fiscall reward for not even a nominal somenecure job.

It’s hash becomes a cop-tag that says “child exploitation” so every journalist that get’s sent what in reality is a politically embarrassing file, is automatically a “kiddy fiddler” or worse. Not just in the eyes of the Police, but Judges, Jury, etc. But in reality the file never ever gets looked at “to be verified” by a human because that is a life sentance or worse crime…

Call it “Justice 2021 style” just for fun, but we can all see it happening or something very similar in the very near future…