'Moore's Revenge' is upon us and will make the world weird Earlier this year I lamented the inevitable death of Moore's Law - crushed between process node failures and exploits attacking execution efficiencies. Yet that top line failure of Moore's Law hides the fact that chips in general are now cheap. So cheap that the cost of making a device "smart" – whether that means, aware, …
Sounds very much like my sysadmin work on Solaris of twenty-odd years ago. Servers were relatively expensive, so we had multiple applications + support scripts on each server.
Add a few patches and middle-of-the-night quick fixes and you ended up with weird behaviour caused by things interacting in unexpected ways.
Gut-feeling got you through at least as often as linear thinking.
Whilst I agree that we're probably going to see increasing chaos due to unforeseen interactions between growing numbers of interconnected systems, I don't think that the problems will be due simply to the number of systems that need to talk to each other, which is what the article seems to suggest. Rather, I think that it will be due to how those systems talk to each other, and I think that the issue here is Deep Integration.
Many cases, Deep Integration is necessary for performance but Deep Integration is also intimate and therefore fragile, and it's because of that fragility that we get so many unexpected problems (fonts causing security issues or, as the article suggests, your car failing to start because your dishwasher is throwing a tantrum).
The article misses a trick though - it was only a day or two ago that another Reg article pointed out that one of the problems with statistically based AI, which is all we have atm, is that we can't really explain any particular output they produce.
When we combine these two factors, I think increasing chaos and unreliability in the relatively short-term future is a certainty.
As to which systems we should be most concerned about - well that's probably a toss-up between the military and food supply. The danger from unexpected behavior by military systems is obvious but with the vast majority of people in the western world only able to get food by buying it i.e. they can't produce their own food, a breakdown in the food supply chains would very quickly result in hundreds of millions of people starving.
This post has been deleted by its author
M. Pesce» We may think the world is weird today, but once hundreds of billions of marginally intelligent and minimally autonomous systems start to have a go, that weirdness will begin to arc upwards exponentially.
What comes to mind here is the hive-mind of a mob, and how a mob behaves differently to the individual. It doesn't inspire hope.
Mobs are only difficult to deal with because we care somewhat about the individuals that constitute them. When a mob is a plague of vermin or of useful devices gone bad, then we shan't actually care all that much about the individual constituent members of the mob other than to rescue something useful from the chaos.
At this point, several solutions suggest themselves. Firstly, stop the chaos replicating and spreading. Secondly, contain it as much as possible. Thirdly, either rely on clean-up systems already present, or introduce them. Fourthly, build some sort of predatory system which actively hunts and destroys rogue devices (defined as devices which do not have the current friendly key).
Basically, we're ecosystem-building, with a hint of immunology built in.
Remember Pratchett's Law for Mobs: "The IQ of a mob is the IQ of its most stupid member divided by the number of mobsters." Extending this well-known fact to the swarm of IoT widgets with limited intelligence to begin with does not bode well. They could make Sirius Cybernetics' GPP feature look like a pretty neat idea (like digital watches).
I'll get me coat.
Doffs hat (grey Tilley once more) to the late great Terry Pratchett and Douglas Adams.
Many years ago, whilst somewhat inebriated, myself and a group of friends came up with a theory about intelligence. We postulated that intelligence is a function of the number of people in a group. We couldn't quite decide whether it was similar to resistors in parallel or a 1/n function (we were a large-ish group). Either way given that the average IQ is 100, the IQ of a full 20,000 seat football stadium is not even up to the IQ of an Amoeba. Personally I now believe that the IQ of groups simply falls to the level of the lowest IQ in the group (Lowest Common Denominator Theory). None of these bode well for the connected world, either now or in the future.
Weirdness from complexity. the example of a blue tooth driver mucking about with Office. Back to the original Unix design philosophy. Each tool or code does one thing onl, fails safely (ie does not crash anything else) and is able to be used in a pipeline. Like the way shared libraries were supposed to provide standard capabilities sharable among programs with standard APIs. INSHO, too much crap trying to do un-needed much. As for the rest of the connected dust, get a bigger vacuum cleaner. The bandwidth available to most of us is already too choked by rubbish packets.
As for complex systems, emergent properties work both ways. Unexpected capabilities, usually expressed as weird and unwanted behavior. Been there done that. (fixed it that is).
Article misses a couple more critical points:
1) It's stated as a given that 'smart' devices will become omni-present, even though for many devices, the possible gains to be made are infinitesimal compared to the cost of data volumes, energy use, privacy, maintenance, and cleanup of the inevitable f**kups. Sadly, though, I think taking it as a given is appropriate since marketing stuff they don't need to people who don't know better is now standard.
2) "The focus on security will produce new costs for businesses - and it will be on IT to ensure those costs don't exceed the benefits...". No, that's not what will happen. What will happen (and we know because we're already seeing it) is that business ensures that the costs don't exceed the benefits by cutting the costs and one of the first places costs will be cut is security.
How do you lose all that information and 1) no one cares 2) they got away with it???
Because no-one in charge cares as long as they can have the next swimming pool while mumbling inanities while receiving money.
> There has been nonstop war since that 9/11 weirdness, no-one cares. In fact, rolling up to the next with glee.
> There has been money printing to a degree unprecedented in human history, no-one cares.
> More and more money and power goes to two ME countries with very strong political influence, no-one cares.
> Europe is manoeuvered into a confrontation with a very fat trading partner on thge say-so of publicly unwell people, no-one cares.
> Population replacement is foisted on countries that do not need, want or are even able manage it, no-one cares. Actually that will probably put the kibosh on Moore's Revenge before long.
There's always big talk on big corps brainwashing people.
...
But haven't they already gotten away with it?
Sure there's an uproar every now and again, but these uproars are usually delegated to small communitues and generally never reach a mainstream audience. What happens when these communities die out? What happens when the ideal of the strong, independent thinker is gone?
Yes, as the popularity of shaving and not smelling like a human attests. Purely for the sake of selling razors and perfumes, our civilization has been brainwashed into thinking that natural pubic hair and the natural smell of a human are evil. I'm still not sure on the score for only females can have long fingernails, perhaps the sellers of fake nails and nail clippers split the market between them?
I think the ideal of the strong independent thinker is long gone.
I see two things going wrong with the Unix philosophy. One, you cannot assume the thing that's doing the one thing is doing it right, and this can result in chain reactions where the next program falls over and so on, leaving a tangled mess with no real assurance one can find the root cause of the problem. Just because it's doing just ONE thing doesn't necessarily mean it's more likely to do it RIGHT (which is a different definition from doing it WELL).
And doing it right leads to the other issue: specifications. People ask one thing but really need another but don't realize it, so some mind-reading is expected so that people get what they REALLY need instead of what they THINK they need only to learn too late it's the wrong thing.
Re: 'I see two things going wrong with the Unix philosophy...':
a) Nothing is perfect, but the KISS principle has generally proved a useful way of mitigating that inconvenient fact.
b) Suggest a better alternative default than 'do one thing and do it well'.
"b) Suggest a better alternative default than 'do one thing and do it well'."
Given where we are.. Moan, quote memes, copy paste the same crap from the last similarly titled story... And do bugger all to change anything. While warning of dire unforeseen consequences should anything ever be done.
"a) Nothing is perfect, but the KISS principle has generally proved a useful way of mitigating that inconvenient fact."
But it ignores two other unfortunate facts of life: the existence of necessary complexity and the nature desire of humans for black-and-white answers in a world with infinite shades of gray. IOW, sometimes the problem at hand has no practical solution (and it's also not as easy to prove it as Turing's Halting Problem proof), yet few are willing to propose to the customer, "Sir, you can't get there from here."
"b) Suggest a better alternative default than 'do one thing and do it well'."
OK, how about "Don't Trust Anything"? The main problem with modern computing is that you can't really trust anything: not the user, not the process before you, and not even the process after you. Things CAN and DO break, and not always for obvious reasons. We're even reaching the point where a "Hello World" can break something serious. Assume that the process before you lied to you about your available resources while the one after you will probably misunderstand you. This Brave New World of computing is probably going to require a total rethink on how we approach solutions: may even force a retreat back to the days where one cannot assume much and may need to do as much as possible with as little as possible (especially with regards to external resources).
"Sorry. What we meant was 'A better alternative that actually allows you to get things done.....'"
It is the only logical way going forward. If you can't operate with the Sword of Damocles over your head, you shouldn't even try. THAT's where we stand now, and we just have to deal with it. Otherwise, we can just throw up our hands and go back to pencil and paper.
The problem is, a lot of those chips are being put into devices that usually last "a lifetime", or at least they used to last a decade or two. Due to the use of chips and being "online", the devices are now dangerous to use after a couple of years, because they no longer get security updates.
That means that these "cheap" chips are causing more landfill or early obsolecense of devices, even though the devices themselves should last much longer... A fridge that costs "real" money, without intelligence, but designed to last 20 years may cost more today, but over its lifetime, it will be cheaper than an intelligent fridge, because it doesn't need replacing every couple of years, because it no longer gets updates.
The same is true of most "smart" devices, the smart functionality doesn't generally bring anything really useful to the table, but makes the device "dangerous" after a couple of years of operation, because it no longer gets security updates. I'll be sticking to non-smart devices for most tasks, thanks all the same.
Like he says , as long as it can carry on without the online bullshit.
Does this fridge really exist? Its become like the standard iot joke. Its like on of those urban myths that are repeated so often most people swear its true. like ... um ... well these:
https://www.rd.com/culture/false-facts-everyone-believes/
Ive just put my phone down on my desk . Does that mean I've created the smartdesk?
No? how about if I tape it down?
I can now do all sorts of things with my smart-desk tm
Stream movies
arrange meetings
play spectrum games
go shopping
book holidays
buy drugs on the dark web
"Like he says , as long as it can carry on without the online bullshit."
So.. like every EXISTING appliance. Smart TVs, despite the assertions of being required to connect all the time "any day now", can still be used for all non internetty stuff with no connection.
"Does this fridge really exist? Its become like the standard iot joke. Its like on of those urban myths that are repeated so often most people swear its true. like ... um ... well these:
https://www.rd.com/culture/false-facts-everyone-believes/"
Yes and no.
It exists in the sense you can get a fridge connected to the internet.
But only in the sense that it is fitted with a pretty basic tablet built into the door. Still keeps stuff cold even when disconnected.
My favorite is the internet of imaginary things oven, that for some reason is unique among similar fitted kitchen appliances, in that it can get hotter than can be contained in it's heat proof enclosure, but is also for some inexplicable reason, fitted with the innards of a pottery kiln, that some hacker is going to be able to initiate.
5 years is nothing like the acceptable lifetime of a TV or fridge, imho.
It may not be the "acceptable" lifetime, but my telly is away for a warranty repair at 4 years old and it seems that they're unable to find a replacement screen panel for it.
I like the telly and I don't want to replace it, but that may be the way it's going here...
(My old Panny plasma screen lasted >11 years before it went kaput. The replacement LG went for 14 months and wound up as landfill when it croaked - cheaper to replace.)
"The replacement LG went for 14 months and wound up as landfill when it croaked - cheaper to replace."
As in cheaper for your time I presume, since it should be free to replace under warranty. Unless I've missed something.
If the companies are going to design consumer goods to be not very durable, and avoid having spare parts available, then surely there is going to be a tipping point where stuff starts breaking under warranty. Which still may be cheaper for the company to replace rather than repair.
"5 years is nothing like the acceptable lifetime of a TV or fridge, imho.:"
Bullseye! No more pesky consumers using a product forever or until it wears out. Now it will simply stop working or become a "HACK ME" sign if you don't dutifully ride the upgrade cycle when they want you to. Companies will be freed of having to support old products, claiming "too hard, legacy" and such.
There's a war on Ownership of Lasting Products, and IOT is a major front in that attack. They want us to either buy the same thing again and again, or keep paying for the same one as long as we have it.
5 years is nothing like the acceptable lifetime of a TV or fridge, imho
And extending the argument a bit, or a car.
We have cars chock full of built in computers, including GPS etc. I have a car that is nearly 30 years old and still runs well, most of the electrics still work. It was made by a company in the VW group. Will VW guarantee that their latest in-car GPS and computer-controlled aircon will work in 30 years?
I have no problem with smart TVs, Fridges etc as long as either it the manufacturer offer security updates for say 5 years after purchase or that any online connectivity can be disabled and the item still be usable.
Security update for a fridge? You mean something like a new padlock on the door?
WTF is your fridge doing on the net? Downloading your meat & veggies? Last I heard you are supposed to eat those not stream them.
This post has been deleted by its author
The same is true of most "smart" devices, the smart functionality doesn't generally bring anything really useful to the table
That depends entirely on your perspective, i.e. whether you're customer or product. One "weirdness" that has emerged quite early on in this process is the blurring of the lines between the two.
In the meantime, I have reached the stage where I can, for example, turn off the garage lights, which I have left on and forgotten about until I'm in bed, by stabbing a button in a browser, open the garage door from within the car when it's pelting down and even schedule the immersion heater for an hour before I get home to allow me a nice soak to wash off the stink of betrayal and incompetence from all around. None of this was obtained from a vendor, though, so it serves me. The other way around is becoming increasingly common - and not just in Soviet Russia, as the meme would have you believe.
In my IT career it seemed that many of the people doing the development/support didn't show any aptitude for understanding complex timing interactions.
Is the ability to multi-track interacting time lines in your mind unusual - or just "too hard"?
One thing I notice these days is that people walking in the street generally do not think ahead. On an intersecting course with other people they only take avoidance action at the last second - rather than predicting a smoother path earlier.
In sailing you always keep course and one predefined party averts collision. Entering motorways works the same way. Problem with walking is we have not decided who moves. We have decided which side to cross which will usually suffice. Occasionally you do a little hopscotch with peeps that have forgotten the rules.
Its interesting how one person's logical thought process produces different results to another.
"In sailing you always keep course and one predefined party averts collision."
A head-on collision course has no single predetermined party. Both have to change course in the same predictable way. IIRC aircraft have a similar fixed evasion procedure in that sort of case.
"Steam gives way to sail" presumably no longer applies if the "steam" is an oil/container ship - even if the "sail" is bigger than a dinghy.
When a pedestrian head-on interaction is likely to force one person very close to the road - then I try to manipulate the passing positions early on. Whichever person is nearest the road should be facing the oncoming traffic - so that they can see any danger to themselves.
"Port tack gives way to starboard tack: If two sailboats are approaching each other and the wind is on a different side of each boat, then sailing rules are that the sailboat which has the wind on the port side must always give right of way to the other." This could apply to a head on scenario, and there would be a clear right of way.
If a motor vessel is experiencing some kind of difficulty restricting its maneuverability, it is given right of way.
a restriction like "being a 550,000 DWT oil tanker" presumably counts
I remember a London-based friend explaining right-of-way on roundabouts and when lane-changing in the Smoke as being that the vehicle with the cheapest paintwork has right of way. i.e. if you're driving a Ferrari don't argue with a clapped-out old Cortina.
"Its interesting how one person's logical thought process produces different results to another."
I'm a Brit. I live in France.
It's pretty much burned into my brain to move more to the left to get out of the way (just like when crossing a road I instinctively look to the right first).
For the Frenchies, they move to the right to get out of the way.
Oops.
With the increasing wages and work safety laws in many of the Asian countries it remains to be seen if chips remain this cheap. Many Asian fabs seem to work on the principle that it's fine if people work 14 hour shifts, and it's okay to lose a few people to HF poisoning every now and then. Once that starts changing (and it is already, albeit slowly) it remains to be seen if chips can continue to be made as cheaply as they are now.
Explain since fewer meatbags tend to mean fewer germs and other potential contaminants that Murphy dictates WILL get out of the clean suits eventually. And you can passivate many types of machienry to make it less reactive to the environment.
a lot of chip makers aren't in S.E. asia. just pointing that out.
But yeah, there are work safety and environmental issues in S.E. Asia, with China being high up on that list. They've been grossly polluting their cities, not just 'the occasional violation' as it would seem. fortunately the western world has made such irresponsibility illegal. Nobody wants dirty water/air. And don't try telling me Republicans are FOR it, either. (That's just FUD).
Some of the worst examples of employee exploitation, dangerous working conditions, and environmental pollution are in a COMMUNIST country. Think about it.
[A strong labor union or two, as well as some noisy environmental wacko groups, might actually do them some good... I hate to admit]
Still, the ONE thing that will render nearly all of that irrelevant will be automated chip factories in the USA and EU. Then the price can be low WITHOUT throwing a "mass of low wage bodies" at the manufacturing side. Instead you'll have a handful of technically competent well paid people managing the robots, and it would make more sense to have these factories "over here" for cost, transportation, and potentially political reasons.
Cheap chips will remain. The way their made, not so much.
"Nobody wants dirty water/air. And don't try telling me Republicans are FOR it, either. (That's just FUD)."
Sure they do. They just don't want it in their backyards. The other guy? Let them rot. And that won't change even with automation since there are various strings attached to the whole manufacturing process that won't change (because of basic physics and chemistry) barring some unforeseen mining, refining, and/or manufacturing breakthroughs.
...[A strong labor union or two, as well as some noisy environmental wacko groups, might actually do them some good... I hate to admit]...
Such groups don't last long in a country where the penalty for causing a fuss (also known as state sabotage or disagreeing with the Dear Leader) is execution...
"Such groups don't last long in a country where the penalty for causing a fuss (also known as state sabotage or disagreeing with the Dear Leader) is execution..."
Depends. If the group is large enough and/or has enough support, they could end up starting a popular uprising, when it's Dear Leader whose head ends up on the chopping block...
BRISTOL CRIME NEWS
ILLEGAL BAGEL TOASTING
A toaster has told this newspaper that an attempt was made to toast a bagel by an unauthorised person. The toaster did not recognise the (human) toaster, but was able to identify the person from the public face database. The toaster phoned the police to complain, and then told the human "I'm sorry Dave I can't do that".
RAILWAY STATION TOILET INCIDENT
Yesterday a toilet cistern at the main station incorrectly identified a customer as a suspected terrorist when it matched the customer's face with police information about wanted persons. The senior toilet cistern immediately locked all toilet seats, all cublicle doors, and the main door to the toilet facility. The facility was fitted with equipment supplied by a major technology company, which refused to accept responsibility, and refused to pay for the clean up. A spokesman told this paper "S**t happens".
Butt how would this work? Do you mean by their arse-print? Or, by poaching from the technology that can identify an individual by their gait, the seat could identify the patron by, erm, "trajectory"? And you could probably infer the worst if the recognition database was called Starfish...
It could identify users by sifting through the DNA in their sh*t, like the ol' Google Toilet...
Hard to believe that's from way back in 2009 if not earlier. Last decade's parody is next decade's grim reality.
He apparently he wrote about circuit densities. That's a rather significant difference. BTW you can turn that into speed if you manage to use parallel processing. In order to do that in any meaningfull way you need to ditch the shared memory concept as it doesn't scale.
It could be that we see a revival of the Transputer concept, where we have small simple processors with their own RAM connected via a high speed bus.
"/me thinks: GPUs (sorta already doing that, for some things)"
I'm not sure, as far as I know they still share the same memory. So they are nice if you have independent processes. I'm not sure how efficient those are for pipelines of different processes chained together.
1. Moore's Law is about complexity, that the number of transistors doubles every two years. Nothing in there about performance, at all. Over the years many have mistakenly tried to make that link, and been corrected for it. Add one more. Further, given Meltdown / Spectre, expect future chips to get much more complex, with both added security and attempts to improve performance to replace that lost by the removal of speculative execution and/or the increases in security. So "Moore's Revenge" could have been about this upcoming transistor count increase boom on the horizon, and been a factually correct concept. Sadly, it was not.
2. "Moore's Revenge" - as it was expressed - is just a rehash of (very) old complaints about first smartphones not having firewall, antivirus, etc. for proper security, then IoT leaping into the same failing with willful blindness. By today's standards it is an ancient truth that when you advance a device to be as capable/complex as computer, then you should protect that device with the same measures as you do a computer. Old news. That no one actually takes that security seriously is a sad testament to today's society, but has nothing to do with Moore. So if anything, this would be more aptly named "Orwell's Revenge" as soon everything around you could be spying on you (or otherwise used against you) and generally was allowed to happen by the ignorance and docility of the masses.
But what do I know? I'm just a commentard.
Quite.
Things that require internet connectivity require internet connectivity. A web browser is just a boring white square without.
But things that require internet connectivity "just because bullshit waffle FUD fudge"? No thanks.
Seriously. No.
The only way you'll tempt me with a "smart fridge" (etc - WhyTF does a fridge need to be online?!?) is if the controller board is documented, schematics are available, and the FULL sources are available on GitHub (or, at the very least, a tarball) so that it can be updated and expanded long after the manufacturer has lost interest (which is usually some point before purchase).
Otherwise, any "connected" things that I will buy are merely considered toys and are usually for "playing with" (in a manner of "playing" that involves screwdrivers and USB serial ports).
"Things that require internet connectivity require internet connectivity. A web browser is just a boring white square without."
Makes you wonder, though, what happens when (not if) EVERY version of an appliance available on the market is smart (against your wishes) with no way to remove them at any price, the stuff is tamper-resistant (break it, break the appliance AND void the warranty because you tampered with it, giving them the excuse of end-user malfeasance), and all the old ones have gotten scrapped due to environmental regulations (introduce a new refrigerant mandate at the same time to cover their kiesters).
IETF's YANG and related encodings may help here I.e. have a standard data modelling language, management protocols, and encodings for managing the plethora of IOT devices.
YANG - RFC 7950
NMDA - RFC 8342
CBOR encoding of YANG - draft-ietf-core-yang-cbor
CoAP Mgmt protocol - draft-ietf-core-comi-02
Alternatively, then could all just use JSON and then everything can be stuck together in an adhoc way using scripting.
They really don't like the old "Man's home is his castle." (forgive the anachronistic sexism). They: gov'ts, corps. Coming soon to a life near yours: Chez pas rien jamais aucunement Château (forgive my purposefully bad French). Now, let's discuss your culpabilité. Let me tell you how I may enhance your life today. No need for pardon, we've got you.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
