Clive Robinson • January 1, 2018 6:47 PM

@ Rachel, Wesley Parish,

I note from your other recent comment you are in New Zealand not Australia as I thought.

Can I say “me too”? 🙂

I must admit I am curious about Kim Dotcom. It’s very clear he is a “marked man” from not just the US Gov, but parts of the NZ Gov as well, seeing as how he has upset their political system. After all their sending in a lawyer into court –to fight for the extradition– to not just lie[1] but grandstand about it as well for the Newspapers does not bode well for justice in NZ…

Further if what is reported is correct to carry on with the pretence that Kim commited a criminal act thus could be extradited is shody at best. Prof Lawrence Lessing showed it was not a criminal act even in the US, It would appear that the Hong Kong judiciary –where the alleged crimes were supposadly committed– agrees as well, and a NZ court has made it clear that he can not be extradited on that either… kind of nails the lid on that asspect.

The fact that from what has been reported all the other charges stem from the supposed copyright infringment being a criminal act you would have thought showing it was not a criminal act would make it all a moot point.

As for,

I note he recently received a multimillion dollar award for being ‘swatted’.

I was under the impression it was paid under a confidentiality agreement. From what I’ve read two other people at the birthday party have already received settlements above half a million dollars, and Kim Dotcom being the householder and most prominent would be expected to receive a commensurately higher award.

Of course it does not help the police case when they make statments about shotguns to make it sound like their behaviour was justified and that Kim Dotcom was a dangerous fugitive with a gun in easy reach. When in fact he had simply run away from “terrorising” people who had failed to properly identify themselves… Again piecing together from news snipits it appears the shotgum in question was apparently not readily to hand but in a locked gun cabinate and from what has been said Kim did not have any keys on him. Also apparently the gun belonged to a security consultant dealing with personal protection (AKA body guard).

Oh and their appears to be a major discrepency over the FBI’s role in the entire event. If the NZ police are to be believed the FBI took no active part in the event… However it appears from court documents they did take an active part and ended up illegaly stealing copies of hard drives etc. Apparently the warrant had been quite deliberatly made over inclusive to enable such activity and an NZ judge was most unhappy about it. So that also adds “fruit from the poisoned vine” issues to the case as well…

Then there was the little FBI/DoJ stunt over Kim’s defence paying for “expert opinion” outside of NZ they made it clear that they would seize any payment which is a prima facie case of “striping of rights”. Which is why Prof Lessing did it pro bono.

If the speculation is correct then President Obama not just pushed the case forwards but instigated it “for friends” in Hollywood and Silicon Valley which is actually not realy that unlikely, bassed on his other behaviours.

Thus it can be seen that as Obama is no longer in office “the driver is absent from the throttle” and there is no “deadmans” to bring the runaway to a halt. Thus there are now careers on the line at both the FBI and DoJ which means common sense is not going to get a look in and more money and resources will be thrown wastefully into the case. Hence it’s become a war of atrition with the US trying to “starve him out” of NZ.

But… There is another issue, Kim Dotcom claims to have knowledge that the DNC hack was as many suspected an inside job. What the truth of of this is unknown currently, but Kim does have “previous” on making claims that his evidence does not support (Warner Bros and allegedly faked Email).

If nothing else the whole process makes for occasional entertainment which unfortunatly also brings in his Ex-wife and Five children needlessly (as I understand it the kids are NZ citizens by birth and thus should have protection under NZ law).

[1] This was over what was apparently a so incompetently translated illegal recorded phone conversation that was so benificial to the lawyer and so prejudicial to the defence that you have to question if it was incompitence or malicious…

Clive Robinson • January 2, 2018 6:29 AM

@ BS PM BS,

Now I wait with bated breath for the AI faecal recognition software that correctly identifies

Is the “bated breath” due to the excitment or holding your nose at the test subjects (we have a cople of “the great unwashed” in London and they do tend to smell of dog poop and rotting feet even when you are upwind of them).

But speaking of scanning things I think it was the third episode of Futurama “A Fish full of Dollars” where Fry goes into the Big Apple Bank and the cashier says “We don’t have your retinal or rectal scans on file… Do you remember your PIN?” it still makes me whince at the thought of it[1].

The last paragraph of my reply to Danial Wood in Aug 2010 still applies though,

https://www.schneier.com/blog/archives/2010/08/skeletal_identi.html#c455853

[1] There was an opening “advert” in a later episode by a drain cleaning robot, with fast rotating hands, where his chirpy drain message ended with a “barbershop” sing of “we can also help with an impacted bowel”… If memory serves it was the episode with the “aqua-exercise pool” with a mothers to be session that Dr Zoidberg interupted in full mating mode and causes a panic which as he leaves is accompanied by a sound track of ploping noises and baby first cries as it becomes a mass birthing pool.

The reason I remember these is not because I’m an avid fan, but my then young son became one and he would watch Futurama over and over and over… Oh and for some reason the episode that first got him hooked was first played on UK terrstrial TV one Xmas morning, and his mother who is a nurse laughed like a hollow drain with tears of mirth at both scenes…

Clive Robinson • January 2, 2018 1:33 PM

@ Anon,

An as yet undisclosed problem is apparently the source of such a significant security bug

Yup it’s real and it’s not software solvable, only partial mitigation is going to work, and then probably not for long based on previous research time lines…

I read this paper whilst convalecing away from home,

https://arxiv.org/pdf/1710.00551

And it opens a whole world of nightmares for those that know how to look at things even slightly hinky…

To understand it the first thing you have to get firmly fixed in your head is the computing stack model from the programmer/user accessable Instruction Set Architecture or ISA down to the transistor level and some of the basic physics below that…

This is territory few ever venture into these days and your chance of meeting someone below fourty who has worked down there during their working career is slim. It just so happens that having been around for a long time 😉 I worked on the designs of bit slice computers where the “CPU” is something you can see as individual chip parts (see 74S181[1] and AMD 2900 and later family[2])

Without going into all the details, what the programer or user actually sees is the ISA which is a long way above the registers and ALU you might get taught at school. At the register level of the stack it uses a Register Transfer Language or RTL CPU hardware specific programing language. In between is the Instruction Decode and pipeline control where the the ASM to RTL microcode exists and things like instruction look ahead, and some cache control are also found. But importantly there is a lot to the side of the ALU that rarely gets talked about and a whole load sat underneath that likewise does not get talked about like Dynamic Memory Access (DMA) etc used for I/O and other low level system related events.

One of the things that sits below most of it is memory, which is mostly Dynamic RAM[3] of some form these days. Simplistically DRAM memory bits are a FET used as a switch to multiplex a large capacitence that stores the individual bit information as a charge onto an internal data line.

By large capacitence I mean with respect to the normal gate capacitence of a similar sized FET. In practice the capacitance is directly related to the area it takes up on the chip and is thus made as small as is possible to get as many memory cells on a chip as possible. It is the size of this capacitor where all the problems start, and has been known about since the first multiplexed refreshed DRAM chip the Mostek MK4096 4 Kbit DRAM (which was engineer weekly wages priced back in the mid 1970s).

The problem with DRAM is that capacitors discharge over a period of time based on the effective resistance across them. Thus in one CR “time constant” period the charge will drop to around 2/3rds of what it is and be virtually gone by 5CR. Thus the refresh trick is required to correct this.

Essentially refresh is where you put/take charge back into/out of the capacitor. There are two basic ways you can do this, “slow and accurate” by measuring the voltage then pulling the capacitor up or down as appropriate, or the “fast and dirty” way. In practice it’s a bit of both which makes ubderstanding it harder than it should be. So simplistically If the capacitor and effective resistance values are predictable then you could just dump 1/3rd of the charge into the capacitor “sight unseen” as this would top the 2/3rds up to the full value and only pull a zero up one third of the way. Thus if the “data sense amplifiers” are set to read at say 3/6ths of the full charge value then you will get the correct data reading. In practical implimentations things are rather more complex and use various tricks such as using the self capacitance of the internal data multiplex traces to store a 3/6ths comparison value via the “pre-charge” phase. Thus there is often considerable “slop” in the design which gives attackers potential wriggle room just as it does in physical security like lock picking.

It’s the slop in the “pre-charge” and feedback sense amps that alows the attacker to force the charge on a memory cell capacitor that RowHamner uses. It’s inherent in the design of DRAM which makes it a problem untill such time it is ever fixed. It’s an issue the much faster but larger and more current hungry Statiic RAM does not have.

When I started playing with DRAM and Z80 CPU chips the MK4116 had got a lot cheaper due to the introduction of the 4164 64Kbit DRAM chip which cut PCB real estate dramatically, thus Z80 CP/M PCBs were as small as a couple of playing cards. However 8bit computers like the 6502 and 8080/Z80 had a problem, their memory was small (64Kbytes) and making more than one program run on them was a real difficulty due to having to define memory areas for each program as well as task and stack switching from a hardware interupt. In mini computers such as the DEC VAX and through to the main frames such as those from Burroughs, IBM, ICL etc this memory issue had been solved by first a software interupt and a form of segmentation then what we now call a Memory Managment Unit or MMU that gives us what we now call Virtual Memory.

Simplistically what you do is put the equivalant of a second very fast stripped down CPU between the main CPU and the physical or “core” memory. It’s job is to hide the real physical memory address from the CPU so at it’s simplest you could make software physical memory address independent. It does this by “address translation” by simplistically mapping a main CPU virtual address to a real physical memory address. The mapping values are held in one or more regions of core memory labled as “Page Tables”, thus they have to be in a known location in memory, which are often duplicated in cache memory inside the main CPU.

It’s important to note that the MMU is a Turing compleate computer in it’s own right with the “Tape” being page table memory, and that as it sits between the main CPU and the core memory it can do all sorts of things that the programmer/user can not stop or even see, which makes it a nice target for certain types of hacker.

In Intel IAx86 chips because they originally put segmentation in the CPU then much later added an MMU things are way way over complicated and it’s been shown how you can turn the mess into a “virtual CPU” that runs below the CPU ring protection methods thus “owns the kingdom”… To make it worse it probably also alows the SGX security mechanism to be breached as well which is what I’ve been thinking about off and on since first reading the paper.

The only real question is “How do you get at physical memory from the ISA?” to change it. Well in theory you can not because there was no designed method to “reach down” around the protection the security the MMU effectively provides.

However theory and practice do not always align and you have to start thinking about what is there by consequence not just design. Thus that known but not much talked about defect in DRAM in effect opened the possability of a time based side channel around the MMU, and that is where RowHamner works. In essence you overload a defect in the refresh logic for the capacitor in a memory cell and thus change it’s state. In most software etc this would not have much in the way of security implications as the most likely result would be a program crash… However some parts of memory are very critical such as the “Page Table” for the MMU that effectively controls the security settings the MMU provides which with process orientated OSs is most of it…

As long as the DRAM issue exists and it’s fairly fundemental to it’s design the side channel will exist and thus “reach down” attacks from the ISA around the MMU will be possible as long as the attacker can “flip the bits” in the desired place to go further with the attack.

From what is written it appears that the “location” part of the –as yet not publically divulged–attack works by a “Cache Timing Attack”. These have been known about for quite a long time as they are a subset of the broader “time based side channels” that the likes of the UK and US SigInt agencies have exploited since before WWII (ie before GCHQ and the NSA had been thought of let alone chartered). The attacks worked against very high security Super Encryption Machines for teleprinter circuits, that used One Time Tapes such as the systems that predated Ben “Pat” Baylys’s “Rockex” machine[4].

One of the valid complaints about the AES competition was that the process actively encoraged time based side chanbels to be built in by default in most implementations. As the NIST competition was technically directed by the NSA and they were very well aware of time based side channels, why did they alow it to happen… Some say it’s actually what they wanted to happen. Thus if that’s the case they got a good ten years or so out of it and probably still do…

The only temporary solution on the software side people can see is to split out the U-area and K-area page tables from each other which means they are in effect “out of cache” with respect to each other and that is where the big performance hit comes from…

The fun bit is that it appears the attack has been developed in a high level interpreted language that nearly every web browser runs by default… Hence the platform / OS independence of the attack…

[1] https://www.wikipedia.org/wiki/74181

[2] https://en.wikichip.org/wiki/amd/am2900

[3] https://en.m.wikipedia.org/wiki/Dynamic_random-access_memory

[4] http://jproc.ca/crypto/rockex.html

Clive Robinson • January 3, 2018 5:52 AM

@ Cassie,

That technique is so ancient it has fluff on it. Of course, that does not mean it is not effective, but I thought that it was basic knowledge that you never, ever, share your source document

Yup it makes even Methuselah look like a “teenage hippy” in comparison. But that is a big part of the problem…

Which brings me to your last point,

I’m sure you know all this, but as ever, I think it is worth re-hashing common knowledge every so often for new readers.

Yes it is a “requirment” to keep saying it almost ad nauseam…

The reason is “society”… As you note with the KGB and many other police state entities and the “disappeared” citizens. Trying to survive in such a society made knowing this knowledge a “life skill” that like crossing the road would kill or severly hurt you if you got it wrong.

As I tell people “All actions have consequences, it’s a matter of degree after the event, which proper planning and response can mitigate”. After all life is only a zero sum game when you die.

The important point though is it is “society” that shapes us and therefore our actions and it can be quite insidious, thus people “sleepwalk into a living nightmare” as past behavious haunt and condemn them in the present…

Some time ago now @Nick P and myself had a chat about how old attacks became new attacks every few years. What sparked it was a variation on the old basic MBR Virus attack. Back in the days of floppies being used as the “data packets” between machines, most vaguely tech savvy users were aware of “Boot Sector” viruses and how they spread. But then Ethernet and PPP became the new kids on the block in the mid 1990’s and people forgot about disks and the MBR. But Flash drives came along and eventually some malware writer either remembered, read in an old book or “heard a war story” and the next thing you know people were getting “smacked six ways to Christmas” by what was in effect an MBR attack…

The point is that contrary to what most people think, technology does not improve linearly, it happens in jumps and rests with a time line like a childs drawing of a flight of steps/stairs. As there are many different technology types at any one time people move from one technology to another as it better suits their needs. They also transfer back when a technology jumps ahead again. Thus you will see data transfer moving from hand carrying a storage device (floppy) to moving down RG58 coax as 10Mb/s Thin net and dial up modems. Then swinging back with CD/R and DVD/RW, then to 100bT and early broadband etc. Eventually it came around to Thumb drives but has gone back to GigaBit and home Fiber etc.

Thus you see this fliping and flopping between two basic classes of technology, but at each flip or flop people either forget or never learn previous security issues.

Arguably these canary/fingerprint systems are a form of stenography that goes back to tattooing slaves heads and hiding scratch marks on wood that is then covered in wax to look like an unused stylus writing block. In more recent times to the likes of Francis Bacon and his binary based codes for writing in a covert channel into plaintext letters. If you have seen a copy of David Khans book you will see a pen and ink landscape picture which has grass or reeds along a bank in variable length pen strokes. If you know morse code you will be able to read a message from those otherwise innocuous pen strokes.

The point is these technology classes and their instances happened and lessons were learned at the time but then forgoton by the majority when a different technology got used.

What @Nick P and myself were supprised at was just how fast this knowledge got lost. OK the MBR trick was nearly two decades, but some others were as little as two or three “tech generations” of around 12-18months each…

Take Reality Winner, she either forgot, never learned or over trusted, so she sent the original print out. I can accept that she might never have been told or over trusted. After all “thinking hinky” is not something most people ever get to do and if they –outside of certain fields of endevor– start asking what are actually valid security questions they will get accused of paranoia or weird / suspicious behaviour by everybody else. You can tell just how bad society in general has got when you see the “See Something Say Something” “Four S” adverts.

What I find totally unforgivable however is the Intercept Journalist that due to compleat and utter incompetence sold her down the river. Investigative Journalism is one of those fields of endevor where “thinking hinky” is not just a requirment it’s an essential way of staying alive…

So if people call me paranoid etc atleast they cannot accuse me of witholding it from them when they make a mistake.

By the way it was not just the KGB keeping typrwriter print outs the FBI did and presumably still does, likewise for photocopiers etc. Whilst it will not get you down directly to an individual machine it will get you a manufacturer and model and even a production run. Which is often enough…

What might appear to fly in the face of the Reality Winner story is my “mantra” that I tell people which is “Paper Paper Never Data” when dealing with legal types as electronic discovery involves a wealth of hidden metadata that can be exploited by them. Atleast with paper you see what you are giving them.

This mantra is because of an assumption which is that the people I tell are basicaly law abiding, and have fallen into the Shark pool that litication is all about. Which is mainly about appearences not facts or what I would call evidence.

It’s when that assumption changes to that of IC/LEO involvment for political reasons over a “sensitive document” that I start telling them about “thinking hinky” and the what to do of OpSec.

In most cases for a “sensitive document” it’s “retype as plain ASCII text remove all but basic punctuation and change words and/or their order, change figures for words etc”. I would then take them through making a clean machine to do it on or how to do it with pieces of paper you then burn and break up the ashes befor flushing. I have other advice for redacting etc. One little box of tricks I realy miss is the Overhead projector and loose heavily creased bed sheet screen to project it on and then photograph with an old style black and white film camera which is slightly out of focus on large grain film stock, then over develop it. You end up with what looks like the old micro-dot photographs with bled out edges on the type face etc.

If you are printing pamphlets / flyers one old trick I’ve never tried is the old oil/bromioil idea but with PCB etching and then silk screen printing. After you etch the PCB you can then “age” it with a sponge with etchant on this will add random surface pitting and all sorts of other issues including under and over etching etc, thus hiding many fingerprint features. My father used to use a Roneo spirt printer for lecture note handouts, and I used to see him make corrections with amongst other things a razor blade and a liquid corrector that used to smell of pears.

Basic OpSec is a quite thoughtful and for your own safety usually introspective task, as in police states (which various western nations are becoming) people will sell you out for the price of a beer.

The funny thing is that what the KGB did with registered typwriters was not stop disident behavior, but kill the economy. Put simply the basic tool of commerce is the spoken and written word, sent quickly and efficiently. Any check or hinderence on those has very significant cost implications thus kills not just profit but competitive behaviour.

What many in the US do not realise is that the USG is justvas stupid. For instance one thing that has amused me in the past is the US fixation on the worthless polygraph. It’s been long known how to beat it, thus rather than see the light and give up the “techno toy” they try to make it illegal for people to tell you how to beat it. The same stupidity applys to TEMPEST techniques, which are based on basic physics and the laws of nature. Then of course is the more modern and equally as detrimentl to the economy Cyber-Offence prioritization over Cyber-defence. Recent malware from leaked NSA cyber-weapons should have been a major red flag wake up call… But no it’s kind of hard to tell which is worse, the US IC/LE organisations or the Australian Prime Minister…

The thing is the Western IC/LE organisation War on Terror and Cyber-Offense policies where and still are a major cancer on the economy of the respective nations. Worse whilst the FiveEyes nations are very vulnerable, other Super Powers and first/second world nations are not so vulnerable and becoming less so as more dirt gets exposed in the US.

Clive Robinson • January 3, 2018 6:49 AM

@ Oh Realy,

But how do you evaluate say an email body in realtime with a hex editor, right?
(you don’t really, do you? I’m impressed either way, but honesty impresses me the most)

I don’t do personal Email any more but yes I do spend a lot of time at the CLI filtering file formates and reading text in a hex editor. Most of which display ASCII text without peoblem on the right hand side of the page, with “.” where non printing charecters are.

The use of the “Mark 1. Eyeball” via a *nix “od” or hex editor is usually the first step with a forensic look at an unknown file format. A look for “magic numbers” is also a part of it but most *nix distributions have a “files” tool to help with that. If those fail a statistical analysis across the file will show encrypted or compressed areas or even byte code or similar. Experience with working on PDF files will give you an idea of just how much of a headache recovering plaintext can be.

If I understand you correctly you want somethin a little more limited which is a program that will read in a good non compressed or encrypted file and find any “text” and then white lists it to remove or even replace those suspicious charecters.

This is kind of a multi step problem with modern file formats because of all the other crap that is in there. When I pull documents back and put them together for people my basic steps back in the 1990s was pull the file onto a *nix box then use “tr” and similar to strip out control charecters and the like, then use “strings” to get most of the document back into a realy rough text file then use that to search and find basic formating information and manually apply it. The user could then ussually load that back into their document editor and reformat it proof it and fill in any blanks etc.

That does not work as well these days for a variaty of reasons the most annoying being Unicode, it is without doubt an abomination, however the likes of Perl and Python scripting languages will deal with it better than most humans. Aside from that the process would be similar, except you would just change the “blacklist” charecters to a special visable char. Having done that you just pull it back into the document processing program and manually remove them etc befor saving to a new file.

The problem is that some document formants alow for embedded programs and fonts, this suggests printing out to a new file format such as Postscript that you can then edit with a specialized tool.

One way might be to print it out as an image file then load that into an OCR program then manually find and correct errors. As I’ve mentioned before a similar technique but actually with real paper works for “energy gap” crossing.

But at the end of the day I find getting everything in seven bit ASCII text files works for me, but then I work with “English” so it’s an “easy option” compared to many languages.

Some modern OCR systems can deal with Chinese / Japanese / Korean standard charecters and convert them into Unicode or equivalent, which on theory means it’s jus the white/black list issues you have to deal with.

I guess that people could write several books on this sort of thing, and I already know of a couple for programers doing “Internationalisation” that precede the major use of Unicode.

Clive Robinson • January 3, 2018 11:46 PM

@ Holy_Cow, All,

A couple people already mentioned it, but this deserves a full post once more details are in.

See Anon’s comment above,

https://www.schneier.com/blog/archives/2017/12/friday_squid_bl_607.html#c6766938

And my reply immediately following it.

As I noted I’d read one of the published papers over the winter holiday period,

And it opens a whole world of nightmares for those that know how to look at things even slightly hinky…

With a further description of the problem being below the ISA in the computing stack, in places few ever get to go in their career these days…

It appears that “in this case” –which is just one of many nightmares opened up– it is both above and below the ALU and Register level in the stack. That is from the latest info[1][2] it now appears to do with the speculative look ahead circuitry that pulls in core memory into the cache by bypassing the ALU and registers and talking directly to memory at a point beyond the security –supposadly– provided by the MMU and Virtual Memory or VM system.

It appears to be a case of “Efficiency-v-Security” that I go on about on the odd occasion or three from time to time 😉

Basically if what has been described more recently[1] is correct, by striving for more efficiency the hardware designers have opened up a time based side channel[4] –via cache hit/miss timing– by which memory a process should not have access to can be enumerated by the process.

This enumeration is possible due to access to very high resolution time refrences to javascript interpreted code used in all modern web browsers[3] which not only makes it platform independent, it also hemorrhages a whole host of other information leaks.

It appears one “fix option” is to lower the resolution of the time information available in the web browser[3]. History tells us that this is a dangerous and problematic fix (think back to the late 1990’s to Smart Cards and powersupply enumeration repeatedly improving and ending up with DPA).

It’s “Dangerous” because by the process of integrating many repeated timings the resolution can be got back (each time you double up the readings you in effect gain one more “bit” of resolution back. Likewise “Problematic” because if you make a significant change to the resolution existing programs will also get broken, possibly irrevocably so, which has liability issues (if you have payed for code that is now irrevocably broken you are going to want your money back or blood or both, which kind of makes Intel the “Insurerer of last resort” where “The buck stops”…[5]).

Whilst what I wrote in my reply to Anon above is just one nightmare case that fitted the then publicly available information, it now appears from new publicly available information to be a different nightmare proplem. Which are just two of the several “thinking hinky” nightmares that paper from last year opened up in my head.

The other thing to note is that the earlier publicly available information indicated it was Intel, AMD and ARM hardware that was effected, the latest only Intel. However my oft made point about the needless over complexity of some modern CPU designs still holds. Which gives rise to the question of “Has Intel gone to far down an evolutionary cul de sac to pull back?” and will that make Intel a real HiTec version of the alleged “saber toothed tiger” evolutionary extinction. Some appear from what is being said to think it will or be close to[5]…

I’m not yet ready to write Intel off but this is sure going to hurt just as the “Pentium Bug” did. If people can remember back, that partially kicked Intel out of an “Efficiency cul de sac” but only a small part of the way.

The problem is like that of Castle builders of the past, you run into diminishing returns problems. Yes you can make a castle taller, but you have to either accept that makes the walls weaker, or build thicker heavier walls that in turn need more massive foundations, which has other knock on effects in terms of cost and insecurity. Or you radically change the way you do things, Skyscrapers only just became possible with steel reinforced concrete. To get higher a radical rethink of how a build was made was required. In effect you got rid of heavy rigid “structural walls” which alowed glass to be used instead that alowed larger floor areas as a bonus, then there was the people moving problem that the original World Trade center resolved with express lifts etc enabling them to be even taller. The point is both changes required a complete reversal out of an architectural cul de sac, which gives you an indication of the shakeup Intel is going to have to go through to survive. It’s not going to be Intel’s first major change, the limits on tranister speed/heat forced them out from the single sequential core CPU into the multi-core parallel world, the next set of changes will be in effect even bigger, and at some point backwards compatability will have to go otherwise they will loose to ARM and for other reasons to AMD. One change might well be that the idea of reprogramable microcode will get extended, by the use of FPGA techniques to make lower level hardware reprograming in the computing stack area between the ISA and ALU/Register level and even below.

[1] https://googleprojectzero.blogspot.com/2018/01/reading-privileged-memory-with-side.html

[2] https://meltdownattack.com/

[3] https://blog.mozilla.org/security/2018/01/03/mitigations-landing-new-class-timing-attack/

[4] As I’ve noted before an engineer “skilled in the art” of both the “problem domain” and the “Secure system design domain” can get limited increases of one type of Efficiency, without reducing security. That is by not opening up timing side channels etc that significantly reduce Security. But they quickly get hit by the law of diminishing returns that makes another areas considerably less efficient. As an example think about “Pipelining” you get some improvment in through put, but not only do you get a disproportionate increase in delay, you also pay heavily with increased circuitry complexity as well as power consumption which makes you hit the “heat death” wall faster than other solutions.

[5] There are stories now starting to appear about the problem being reported to Intel in Nov last year, and that the Intel CEO cashed in a large part of his Intel shares holding also in Nov, thus implying “Insider Trading”…

Clive Robinson • January 4, 2018 6:01 AM

@ hmm,

What did you mean? I don’t get it.

A large number of people believe –rightly or wrongly[1]– that the bed rock of society is the family. Not just a cohabiting family in a house with suifficient rooms but a family that sits together eats together at the same time talks together etc etc thus establishing very strong bonds. A subtext of this view is that children should only leave home to marry and setup a family of their own quickly and have atleast an heir each and a spare preferabley a couple. Thus three to five children with a two to four year age gap[2].

Thus under that conservative religious view point girls are for breeding and boys for minimal wage labour, both bereft of a higher education that drives economic success. The Church gets new blood and 10% of their labours, the business men get low cost drudge labour that can be hired, fired and mistreated in any which way the employer choses, individuals rights get crushed and shaped to that of the church and employer.

And that’s how it was except during major conflict when men were told to be cannon fodder untill the start of the sixties.

We are led to believe that the sixties were all about “Sex Drugs and Rock n Roll” or various hippy communes etc… It was not much of it was “self empowerment” industrialisation and relaxed post war restrictions ment 14year olds with real money in their pocket and a growing high street economy for them to spend it in. There was also plenty of cheap rooms to rebt etc, thus for girls the transition was nolonger “Dolls to marital bed and motherhood”. Yes there was a relaxation in morals but not realy as much as you might think (morals were actually much looser in WWII and i n the Victorian era).

The two “scary things” for the establishment were in cities people stopped going to church did not marry and had less children and labour was actually a shortage thus high wages and even accepting women in the workplace did not cover the shortages.

But a third problem hit in hard. WWII had been a “technology war” and thus the real birth of electronics and industrial technology. That needed a skilled work force that was not there. In the UK the school leaving age was raised to 16 and technology collages and other higher education started expanding quite rapidly, especially slightly later with the polytechnics.

Thus the base of society the together familly was on it’s way out in part in the short term to drive the needs of industry, but for women it ment real “freedom”.

It soon became clear that those wanting early marriage was actually the “males” not the “females”. Desmond Morris’s point is that men are actually inherently social and eat that way (see kabab shops at pub closing for example). But you rarely if ever see people walking down the street sharing fruit let alone vegetables (though that has changed some with mothers and children).

The point which is actually more than a trend when you dig into it is that where they can women are not marrying, they carry on studying and get better jobs and social standing their own personal living space to live as they want which is in effect unencumbered. If they do decide to have children it’s very much their choice of when, how many and with whom.

And yes the diet re meat-v-vegatarian is a strong correlating factor in this.

As for men, when you pull off the macho bravado struting talk, they are the ones sitting on the shelf today in the average and above IQ range.

There are two points arising from this.

1, the population size is falling and the age gap is widening.
2, it’s falling and widening most in the average and upper age range.

From the conservative religious view point the world is now extreamly “anti-social” as they see it. But worse it’s effecting the tax take by reducing it significantly whilst people are living a lot longer. There are significant even more anti-social effects because of this.

So the real question is being increasingly vegetarian a driver, correlation or coincidence? I’ve heard many arguments for each case. Perhaps the oddist is it’s a natural effect population control in high density populations. There is actually a very real cause and effect relation between fertility in women and eating both dairy and meat products so it might not be as odd as it first sounds.

It’s one of those things you have to do your own research on and come to your own conclusions, because there realy is not an ethical way to experiment reliably.

[1] The idea of a “together family” especially with a male “head of the house hold” who sets the thinking and policy of the family is often pushed by those with very conservative views that are often driven by religion.

[2] Again a conservative religious view point bassed on continuous indoctronation into the “raising children for god” view point which makes a religion very wealthy over time. It’s also seen as good for the economy, especially by those who profit by cheap labour (see the history of “The Protestent Work Ethic” to see the propaganda in it and who benifits).

Clive Robinson • January 5, 2018 10:01 PM

@ gd,

I especially appreciate Clive’s insights and hope he doesn’t mind the publicity I have afforded him.

As our some of our antipodean colleagues say “No Worries”.

The important thing as I said is the message not the messenger, as long as it gets out as wide as possible, hopefully people will learn from it, and “whilst climbing up on my shoulders reach up to do better things still”.

Although my shoulders may be getting old, they are still broad, just as long as they don’t pull my beard on the way up, I need it to keep my chest warm 😉