Posts by SPiT

There are perfectly reasons to write a paper on facial recognition of criminals but what it should be about is researching the confounding factors that have given an AI the ability to pick out pictures of criminals. There was a Chinese AI project on this which was shown to have developed the ability to recognise official police photographs of criminals because they included them in the training set. That would be good research as it would support understanding of the impact of the photographer on AI analysis of any pictures.

I'm surprised nobody has mentioned the classic "turn it upside down" treatment. I've lost count of the times I've restored a failing drive to temporary serviceability by simply turning it upside down. Stiction can occur simply because the lubricant has very slowly moved under gravity and a simple gravity reversal can give you full serviceability for at least a few days (never tried it for longer).

What we need now is a class action against Amazon invalidating the Prime contract forcing them to repay everyone who signed up via this method since no valid contract exists (that is pretty much what the ASA verdict means)

Nothing to see here

This is a simple batter sale not a carousel. A carousel requires the goods to pass all the way round the loop (ie back to the original seller). In this case the two companies have battered to sell each other their product and not bothered to negotiate discounts potentially just to boost the sales bonus. There is no malpractice unless this was misrepresented in the accounts or the software battered was then simply discarded. I suspect that we are simply going to go back to the auditors being aware of the transaction and approved the accounts which means nothing improper in relation to the sale unless HPE want to sue the auditors and presumably, when this court action crashes and burns, their appetite for another court action will be somewhat reduced.

HPE have to prove that they were lied to or something illegal happened. They really don't seem to be doing that.

I would imagine that the reality is that the HPE team would like to go on a fishing expedition with these witnesses in cross-examination but can't issue their own witness summons because they have no specific evidence they want from these witnesses. That seems to be the only explanation that holds water and is highly plausible given the state of this trial.

no longer required US components

Can we assume this is the key phrase. I always expected that to be the main trade consequence of Trump's ban. Perhaps he's suddenly realised that unless the whole world bans Huawei then he's just disadvantaging the USA.

Meaning of consultation

There is plenty of established law in the UK around redundancy and consultation. The generally approved interpretation has simply led to there being established timescales for consultation rather than depending on receiving responses. In redundancy situations this often means that both employer and staff are left waiting in limbo for the "consultation" period to elapse before being able to proceed. This is often a situation that nobody likes especially where redundancies are being made in smaller businesses where it could very reasonably and fairly be wrapped up quite quickly.

Without looking into this in further detail I would imagine the situation is simply that the judge considers that sufficient time had passed for consultation and the fact that the worker committee hadn't responded is not relevant. The key reason for this interpretation will almost certainly be to prevent mandatory consultation being used to permanently stall the process.

If you don't like that offshoring policies of companies like Oracle then the answer isn't to rely on details of the redundancy process, it needs to depend on internationalisation of things like the TUPE regulations that basically say you can't make people redundant unless the work they are doing is redundant. What isn't clear to me is why nobody is pursuing a TUPE claim around these redundancies.

I would have thought the situation outside of the court case is that Golden Eye are collecting personal data and therefore are required to be registered and have a statement of how the intend to use that data and the lawful basis on which they are doing so. I don't know what impact this would have on the case but there would definitely be an interaction between the two as their lawful basis would have to be compatible with the request made of Virgin.

My understanding is that under modern English law (so I don't know if it applies in america) is that arbitration might be binding for a breach of contract but because the plaintiffs are claiming a breach of law the contracted terms don't apply. That should mean that this can be converted into a court case and converted to a class action. The mobile companies are actually more exposed because the FCC haven't enforced the law and hence the plaintiffs can pursue that breach of law themselves.

Typical court case

Unfortunately the great mass of evidence being presented is only tangentially relevant to the very simple core question of whether HP made a mistake or Mike Lynch fraudulently misled them. The way that English law works is that everyone has to present every single little piece of evidence that they think is even marginally relevant to whether the sale was fraudulent and if so exactly how much HP was defrauded of. This means weeks and weeks if minor drivel that may effect the level of damages since they aren't allowed to try the simple "is it fraud or not as a separate issue. Having taken part in such a case in the past I'm willing to bet that the judge understands the evidence better than the barristers involved and some of it will sufficiently specialist that they will repeatedly make fools of themselves with plenty of unintended comedy in court.

Has anyone at the Register considered reviewing their articles of association to see if there is anything in there that conflicts with what they are doing. For example "The objects of the Company are to undertake activities, particularly (without limitation) as were formerly set out in the Company’s Memorandum of Association, and to do so for the public benefit." which indicates an enforceable obligation for the Nominet board to consider the public benefit of whatever they decide to do and, since it is covered by standard company law, this is enforceable in the courts. They could be obliged to release their board minutes to prove their discussion of how they decided the current plan was of public benefit.

Another failed demo

My favourite failed demo was many, many, many years ago when system builds took a long time. Our manager had been given 6 weeks notice of having to give a demo but didn't bother to tell anyone. When he did at 3:00 in the afternoon we were 2 hours into a complete system rebuild to incorporate some core system changes which we expected to break things all over the place.

In case anyone is still reading, the regional control centres that the UK government commissioned for the FireControl project (go look it up) had their equipment room EPO buttons next to the magnetic locked exit door AND they didn't have an exit button for the door so it was the only button present. The situation was in fact even worse than that in that their official process at the time I came across it was to allow third party staff into the room with no means of exit and expect them to phone security to get out, a scheme that didn't even work reliably as during the delivery phase the security office wasn't reliably manned. All in all a rather impressive breach of health and safety and lone working arrangements.

Muncipal ISPs

I'm surprised nobody has pointed out that the american public would have legal redress against a municipal ISP under the first amendment since as part of government they are obliged to respect it whereas the private for profit ISPs aren't obliged to respect it. Seems to me that the threat to the first amendment is the exact opposite of what Mr Oh Really is promoting.

Isn't there an obligation for accuracy

See title. By discarding the Windrush data and then maintaining a record of an individual's immigration status which is then, as a consequence of having discarded that data, inaccurate don't we then have a case where the home office holding inaccurate data as a result of their own negligence. Isn't it reasonable to suggest that by discarding the data they have in some way forgone their right to hold other subsequent data on the data subject. I suspect that this question would be extremely unclear in law and would have to depend on action for maladministration rather than just data protection.

For example, this would be similar to me destroying all records of a customer's contract and then when they contacted me for contract fulfilment claiming there was no contract. This would typically end up being treated as fraud with the only defence against criminal liability being a claim of incompetence. The Windrush data was destroyed after the beginnings of a period where the home office had started it's policies that made this data important.

Understanding the real horrible nature of Spectre

For those of you who want to understand the full range of Spectre class faults they come in 2 basic classes.

1) Exploits that use various techniques including speculative execution using branch cache seeding to determine the contents of in process memory which the process is allowed to access. Note that this memory is permitted access according to the processor design model so AMD have a strong defence. The exploit is basically used to break the security model of sandboxes that protect other scripts in the sandbox by preventing the execution of instructions accessing each other's memory. Speculative execution breaks this.

2) Exploits that use various techniques to exploit the basic shared nature of cached information. This applies to the in processor cached stuff (branch predictor, return cache, level 1 cache) and the across chip cached stuff (higher levels of cache). This caches information leaks and this is not preventable without making all cached information private to the process that owns it.

The second class is a serious unfixable problem but it is not obvious how it can be exploited. Just think about it, the only real fix is no shared cache, every cache line is private to the process that fetched it and the fact that it is present is private as well. This basically means a private cache for every core, no dual thread cores unless both threads of execution belong to the same process and a full cache flush on every context switch.

Do you want a Spectre vulnerable CPU or the performance of a machine from 20 years ago.

In a slightly different approach, this article contains numerous statements that are manifestly untrue and provable so in court. In particular, they can be proven false to the point that no reasonably diligent reporter could reasonably consider them to be true. In England he would have a pretty open and shut case for libel.

I am seriously disappointed in the Register as they normally take a stance of trying to present the truth. I won't join in with the general mudslinging except to say that I would love to find some anti-Damore article that actually presents actual evidence as after many hours of searching I have been unable to find even a single one. Since his memo is actually available it would be perfectly feasible for someone to point out any of his statements that is, in context, incorrect but nobody has bothered to do so.

All those attacking him are doing a serious disservice to their supposed objective of reducing discrimination as they are insisting that as a society we must ignore the real problems and focus on issues which are either largely resolved or barely ever existed. The current action against Google for discrimination against women is a symptom of this problem. If you ignore the real reasons that men end up getting paid more than women then you end up not doing anything about it.

Can anyone explain why we should consider SPECTRE a hardware fault

I continue to be confused about why everyone keeps talking about SPECTRE as if it is a hardware fault. As far as I can make out the issue is that SPECTRE can be used to read the entire readable address space of your own process. The hardware security promise is that you CAN do this. The problem lies with the whole idea of running a sandboxed program within this environment that cannot see the rest of the address space without having a security rules commitment from the CPU manufacturer. The only sensible way to have safe sandboxing is to have a hardware promise behind it. The obvious solution is for all scripting / sanboxing solutions to use essentially the same solution as for meltdown, make sure that the only mapped memory in the process environment as that which is allowed with the benefit that if you don't have a meltdown problem you can use privilege flagging to protect it.

I appreciate that there are issues around how virtualisation works but the idea that Javascript, for example, wasn't massively exposed to sandbox breaches anyway is madness.

Total Load

I'm more concerned about the total load in 20 years time when we are all (most) supposed to have switched. If my family switched to all electric vehicles then our annual electricity consumption would roughly double. This is a severe threat to all of the electricity delivery infrastructure and hence needs major national development of the entire network as well as the power stations. If on a national basis we can move to EVs and manage recharging to strictly off-peak to even out demand this could all work quite while but that would be a huge undertaking.

On a more local basis my house has a single 60 amp feed and it would be unreasonable to ever run anything more than a single 7Kw charger off that which would be a major issue for multi-vehicle families who want to charge everything over-night.

We are a long, long way from a viable solution but it is possible to identify all the problems and start considering what should be done. The biggest issue is that this isn't really happening.

Could Google discourage them

This simplest way for Google to discourage this behaviour is to implement (and tell everyone) that it downranks pages for each valid DMCA notice but upranks the page for each false DMCA claim. This of course doesn't work because websites would immediately start "false DMCA" attacks on themselves. But is there some page ranking algorithm, based on information available to Google, that could actually deal with this situation.

For example, Google could advise specific vexatious DMCA sources that they have been classed as vexatious and in future their false DMCA claims may result in page rank upgrades for the targeted site. This would turn into a game of cat and mouse between Google and these people but would give them a serious problem in having to expend a great deal more human effort in managing their DMCA attacks whilst Google's processes could be broadly automated.

There is also the teeny weeny detail that Airwave coverage is defined as in cover if the radio works everywhere and the mobile networks define in cover as the phone works somewhere within 100m. I have seen a specific example of a Policeman doing his job involving extensive use of his Airwave radio in a location where mobile coverage (as used to call them in the first place) required walking 100m up the hill (and away from the incident). It remains to be seen how that will resolve itself.

Silver Iodide won't work

The basic flaw in this is still using silver iodide. If they really want it to rain they need something that ios known to seed clouds at much higher temperatures which means they should pursuing research on cloud seeding substances rather than delivery arrangements. Unfortunately they probably need to move to biological agents and I can't see them getting that authorised very easily.

All writs act

The issue in the NY judgement is simply that the FBI have presented an interpretation of the all writs act which permits them to require a third party to do anything that they want as long as the purpose of the action is to aid the FBI in their statutory duty. The magistrate has quite correctly responded by saying that if that is the correct interpretation of the all writs act then the all writs act is unconstitutional and therefore the courts should take a different interpretation or simply refuse on the basis that the all writs act is invalid.

It is that alternative interpretation that is rather difficult. Can you define a general scope definition for permitted writs that includes what the FBI is asking of Apple but excludes outrageous requirements like "Microsoft, we require a complete computer forensics package, please write one for us and hand it over (free of charge)". If the FBI wants to win then they will need to produce a formal justification why the software development work requested is a reasonable request and quite different from my outrageous suggestion above.

The key difference between this and a normal search warrant or subpoena is that it is a request to "do some work" rather than hand over some material evidence.

My favourite sort faux pas

Many, many years ago I had to assist a colleague who had a sort problem. This is in the early days of relational databases on a computer with relatively limited capability. Unfortunately I can't remember all the details (it was about 30 years ago) but he essentially retrieved each record from the database searched for individually to avoid sorting and taking into account the database performance and his program's internal algorithms the execution was order N^4. He requested my assistance because his program worked fine when tested on a 10 record example (although he really should have twigged that taking 2 seconds was rather too long) but when tested on real world data (only 120 records) it just "looped". He did go an fix it when I pointed out that based on his algorithms expected completion time was about 12 hours and it was working just fine. I wish I could remember all the details.

Chernobyl actual accident profile

Just as a correction to all of you who saw other reactors as being as unsafe as the Chernobyl design and having had similar accidents.

The Chernobyl reactor design has a positive void coefficient. This means that when it loses coolant the nuclear reaction accelerates which is really, really not a good idea. ALL western reactors have a negative void coefficient - if they lose coolant and if you leave it alone it shuts down.

Further, the Chernobyl reactor did not have a melt down, it actually went prompt critical and the explosion in the reactor chamber was a major nuclear release rather than a hydrogen explosion. This was all reported in the atomic energy agency final report but for some reason wasn't considered news and hence did not appear in any of the main stream news media at the time. I suspect that the Internet wouldn't let that happen today.

If anyone is still operating a Chernobyl style reactor they are completely insane and seriously need to stop it right now.

I could make further comment but I really can't be bothered to butt heads with all the ignorant trolls, I'll let the rest of you fight it out as you see fit.

The problem is the idiot minority

To be fair almost everyone commenting on this thread has a point but throughout the issue down to the idiot minorities. It is virtually impossible to work out which group of road users has the largest idiot minority. My experience of driving outside London is a series of very clear rules to follow

1) On approach to any sort of left hand bend check behind and make sure that you are NEVER overtaken by a caravan on a left hand bend as most caravan drivers seem to imagine the caravan goes no closer to the kerb than their car does.

2) Always look at drivers in side turnings as cyclists are invisible to some people unless you make eye contact.

3) Listen for vehicles approaching from behind and if they accelerate they are probably about to turn left right in front of you

4) Pedestrians are suicidal idiots but they won't kill you

There are more

Likely First AIs

It is much more likely that the first AIs won't be embodied systems of any sort - not a specific machine or a robot. Also, the first AIs what be the genuine superhuman AI it will be an "alternatively talented" AI. I would speculate that first AIs and in fact the first problem AIs are going to be created by stock traders in an effort to exploit our financial markets (all of them). There is big money to be made, much more economically than in expensive factory automation, and these will be AIs running on whatever hardware happens to be available. Whoever programs them is going program them to "win" without due regard to any safeguards. They won't be very advanced and hence we are likely to get the problem of badly behaved AIs even before we are willing acknowledge that this is what has been created.

This idea will seriously upset the local Fire and Rescue Services. They already have a major issue around dangerous components in cars (gas struts, air bags and electrical supplies). The vehicle will need to be designed so that it is very clear to a rescue worker with cutting tools exactly where they can safely "disassemble" the vehicle to release trapped and possibly seriously injured occupants. The emphasis here, is that they have to know what is safe and what isn't. It isn't enough for there to be a safe approach, it has to be guaranteed safe (else their senior management cannot ask them to do it) and it has to be clearly communicated to the rescue workers in a way that guarantees they will be aware.

If you want to store lots of electrical energy (which they do) then this is going to be seriously difficult to make safe.

Difficult to make a good case against Google

The real issue is that the only reason that sniffing this traffic might be unlawful is if we are claiming that the users of that Wifi had a reasonable expectation of privacy. In a way its the same issue with StreetView. The complaints about street view photography have been principally around the fact that the raised height of the camera allowed it to see things that people would reasonable expect to not be seen from the street - ie over a 6' fence in a street where no double decker buses go past.

So the real question is do the posters who think there should be a significant fine believe that users of unencrypted Wifi have a reasonable expectation of privacy? Personally I don't know if a court would consider that reasonable or not. I wouldn't expect my encrypted traffic to be private.

And, on another note, cracking the encryption on encrypted traffic would be quite another and clear cut offense in the USA which has a law about "bypassing encryption schemes" and could in all likelihood be treated as a criminal matter where you can't just pay a fine.

The reality of global warming

The real problem here is that you are all arguing about the details. In a way Lewis Page is perfectly correct to treat the wide range of global warming fanatics with a degree of skepticism. The vast majority of "information" posted by both sides of the debate is at best meaningless crud but more typically distortion based upon prejudice. What currently gets categorized as global warming science is actually the equivalent of weather forecasting but without the short-term feedback for quality control that weather gives you. You find out about 5 day weather forecasting every 5 days. Attempts to forecast climate change details are going to be very difficult and fraught with error.

What this means is that we don't know exactly what will happen or exactly when. This doesn't mean we know it won't happen, which is the typical anti-warming argument. To use an analogy, we are trying to work out whether it is safe to cross the road in thick fog. We can't see through the fog despite the best attempts of current science but we do have a more general theory that is solidly based that says traffic is coming. The warming fanatics are saying we are about to be run over, the anti-warming fanatics are saying that its safe to cross if you can't see anything coming. They are both wrong but I but caution seems more sensible than the gung -ho 'of course its safe' attitude.

The real global warming science is based on sound Geological analysis and a broad confidence that if we keep pumping CO2 into the atmosphere than it will get a lot warmer and sea level will rise and it will make a pretty big impact on human civilisation.

For a good real issue with the predictive science element of the global warming story the best current example is the realisation that the models haven't correctly considered the impact of a reduced equatorial - polar temperature gradient and its impact on the jet stream and the apparent real consequences already seen for weather in the mid-latitudes. This is a perfect illustration of the weaknesses of current modelling and the desperate need to penetrate that fog and find out what is going to happen.

We really don't know how scared we should be. The "it will never happen" brigade should perhaps be compared with the opposite extreme of some of the earlier predictions from James Lovelock of Gaia hypothesis fame (who has since changed his mind). These represent the extreme positions

So what is the contract

The EULA thing is all very interesting because under English law a contract exists if there is an unambigous offer (software made available) and an unambigous acceptance (ticking the box). However, the unambigousness means it must be clear what the parties actively agreed to and this isn't clear on the consumer's side. Also, even when a contract exists the terms are considered to be whatever the two parties reasonably believed them to be. Supplement this with European consumer legislation which basically says the provider has an obligation to make sure the consumer understands what they are agreeing to and you end up with very little.

The bottom line is that where the agreement is with a consumer (non-commercial user) the fact that Adobe (or whoever) can reasonably determine that the EULA isn't being read the contract can be presumed to be whatever the consumer thought it was - ie yippee, free software - full stop, end of agreement.

So as a home user the likes of Adobe don't have a leg to stand on as far as your obligations are concerned. It might give them some slight protection in terms of liability but I suspect that mostly won't work. If you sued Adobe because a flash install wrecked your computer then their defence is going to have to be the common law defence that the consumer had no reasonable cause to expect Adobe to take liability for their use of free software. The EULA simple backs this up as an "if you insist on looking at the paperwork then its on our side" measure.

Not really anything very new

The implications of special relativity are quite simple - lightspeed is a barrier to going faster than light as transition requires infinite energy. On top of that you might also consider that it can be viewed as a theory about simultaneous events and anything going faster than light maps onto a reversed time line. That is, to an observer, anything moving relative to them faster than light will appear to be moving in time. The end result is that transluminal travel creates time travel causality problems.

Bottom line tends to be that there aren't any rules saying you can't travel faster than light, just rules that say you can't get there. The natural 'speculation' is to imagine quantum tunnelling through the light speed barrier but since it is infinitely high this doesn't make sense either. The whole area has been done to death many years ago.

Have you all missed the point? or have I?

I think the message from this is that something which is accepted as NOT a breach of copyright when carried out with a physical piece of equipment used in the home (recording TV to replay later) is being classed as a breach of copyright when exactly the same activity is carried out via purchasing the service from an online supplier.

To use an analogy, let's speculate that someone has copyright for the recipe for scrammbled eggs and it is generally agreed that you may use it free to cook scrammbled eggs for yourself at home but cafes must pay for the right to do so. Then comes along a service which gives you access to a kitchen on "commercial premise"s to cook your own meals, let's say rented accomodation, and this is then classified as a restaurant rather than an at home activity. Suddenly looks kind of weird doesn't it.