Posts by Doctor Syntax 40413 publicly visible posts • joined 16 Jun 2014
Testing was available before the app. There was no excuse for going straight to self-isolation. The only action to take based on a positive result from a presumptive test is to apply a more definitive test. When telling people to self-isolate they should then have made provision to support those self isolating. If the latter had been done a back of the envelope check on the likely costs would have suggested requiring a test first. In fact I'm surprised the Treasury didn't demand testing on the basis of likely cost to the economy.
The likelihood of this happening was obvious before the scheme was launched. Anyone capable of thinking one step beyond "wouldn't it be a good idea" would have thought about what happens beyond a match being identified. Unfortunately that requirement seems to exclude any politician, or at least any minister and their SPADs.
A match on this basis is nothing more than an indication that a chance for an infection has occurred which is very different from saying than an actual infection has occurred. The likely number of false positives would dictate that the next step should be to organise a test to see it it actually occurred.
If the idea had been handed over to someone capable of working that out for themselves it might have been saved. Instead it was handed over to the person who was even less capable of that than the government.
The two cases are different.
The weaknesses of the ID parade are that people are trying to match a fleeting glimpse with a limited choice of people who are put in front of them. There's also the possibility of suggestion. But it's attempting one in few identifications.
The weakness of AI, at least in mass surveillance, is that it's trying to fit faces from a large database to members of a crowd. It's attempting many to many. Even if it is better than human efforts, which remains to be proven, because it's making many orders of magnitude more comparisons.
From a justice PoV the worry is that it will be believed because of the thought that it's "objective" and numbers can't be wrong despite just about everyone having had experience of a computer letting them down.
I think there's ample evidence that marketroids are very, very bad at working out what people want. Anyone who doesn't block adverts has tails of the irrelevant or no longer relevant ads following them online. Anyone who regularly uses Amazon or other online store has a similar experience with recommendations.
It follows that if they think people want what they don't want they're likely to think they don't want what they do want and consequently products get discontinued or at least becomes more difficult to find. You may be on a mission and prepared to look where the product is now but if it's no longer stocked you're not going to find it.
"The technology doesn't put people in jail, the justice system puts people in jail, and the only way evidence can be used to support a conviction is if that evidence supports that conviction."
If he evidence is from the technology and the technology is faulty then it can indeed result in wrongful convictions.
As someone directly involved in producing evidence one of the most stressful elements in my job was the possibility of being involved in a miscarriage of justice if I got it wrong. It rather seems as if distancing from the actual business of going into court has removed that concern for those developing this sort of technology.
In the Horizon instance the system was producing evidence of crimes were there was none. In more general situations, if there is a genuine crime and someone is wrongfully convicted it also means that the real perpetrator will have got away. That's something that should worry even the keenest advocate of raising conviction rates.
And how do you even know if X is using face scanning?
Not that I really care that much - it's too much trouble to get into my nearest big town. Like so many they're hostile to cars and public transport from my own small village is not good. My best bet if I need to go into town is to drive to another, bigger village, park there and get their better bus service although obviously for the last year plus that's been a non-starter. But when I park there I might as well do any shopping there that's possible there or maybe some other village and for the rest there's online.
For email services I'm quite content to pay for a service and domain name to go with it, not least because it's then easy to issue different addresses to different correspondents and to chop one address if it gets abused.
For syncing between devices there's home storage, in my case NextCloud on a Pi although I assume commercial domestic NAS devices would do that.
As to the net being paid for by advertising, I wonder when advertisers will start to question what value they actually get. Just how many adverts does the world need?
"After that, I will need reasonable alternatives to Google Calendar and Google Docs."
I'm curious as to why people need - or think they need - such services.
Is it because they want to share data between devices?
If so do they never have two devices in the same place at the same time? If they bring devices together then they can, at least in theory, be synchronised. If they can't in practice then look to the vendors for an explanation (or, more likely, an excuse).
Given that my (non-smart) water meter is in what looks like a pretty good Faraday cage I'm still puzzled about how they can read it from a range of several metres. Apart from being underground it's enclosed in a steel tube on top of which, at ground level, there's a hinged cast iron cover. Could the gap between cover lid and frame act as a slot aerial.
Microsoft does not want it called VDI (Virtual Desktop Infrastructure). "We're not shipping anything that's infrastructure. We're providing all that as a back-end service... If you were to classify it, it would be most aligned with DaaS (Desktop as a Service),
Some mighty fine hair splitting there.
"BTW. I imagine that Google's response other than sending a platoon of lawyers to France to argue compliance with EU copyright law will be to stop ever reporting news from French sources."
This seems to be the best option from Google's PoV, giving, as it does, the possibility of letting them pay to have sponsored snippets appear.
The simpler alternative would have been for Xerox to realise that if they took the ideas they had in the Star, reimplemented the hardware using the new-fangled microprocessors and, by that means, brought the price down they could have owned the business market for years if not decades. Unfortunately that was never the Xerox way.
The date is set for the sprint to finish and higher up management expect you to meet it.
Perhaps manglement would understand a car analogy:
Your car is in for service and is to be collected at 4 pm. An hour after it goes in the garage rings up and says "We've discovered a problem with the brakes, The parts won't be here until tomorrow morning. You can have it working at midday tomorrow or collect it at four today as agreed but it won't be fit to drive. Which do you want?"
"If we managed R&D projects like we managed building motorways they'd be a disaster. They'd all under-deliver, late and over budget - just like software projects."
Cross Rail belongs to the same branch of engineering as motorways. Just saying.
NEC made the same promise in November 2020 – but for AWS engineers. ... NEC and Google also struck a deal in 2018 to “better serve Japanese enterprises”.
I'm reminded of an old saying: "once is an accident, twice is a coincidence, three times is getting to be a habit". Or do they believe in "third time lucky"?
It's only relatively recently that the UK got round to issuing campaign medals for the merchant seamen on the Russian convoys.
A classmate at school had lost his father in the Merchant Navy. Just because they weren't shooting anyone they seem to have been largely overlooked.
The solution to this, as in the aviation industry* is a statutory inspectorate. It would need powers and resources for pre-emptive inspection and prosecution.
Unfortunately we have a govt. that keeps bleating about being the best in the world for whatever issue drifts across their minds on the day but usually mean by that an absence of "red tape" when, in fact, the said red tape holding in place the security of the supply chain would actually be more beneficial.
* Withou Boeing-style self-certification.
That's not what's being proposed although it was how I read the headline.
What is being proposed is something akin to what you suggest, except that the large mass is stuffed into it into the satellite chassis instead of the usual electronics. Presumably the idea of doing it this way is that the satellite itself has all the necessary bits to connect to the launch vehicle although a lot of expensive stuff that wouldn't be needed. Why not build a fleet of deflectors containing no more than is required to do the job designed to fit each generation of launch vehicles? It would cut out a lot of arguing and lead time if all that had to be done was commandeer the actual launches.
"well not yet quite for email, I have my suspicions that MSFT are working on that"
Judging by the rest of the comments you've now got a lot of people worried.
"From the support issues I've seen it is the users who get confused, not the system!"
It may or may not be the users who are getting confused but if they are it appears to be the system that's confusing them. The role of system designers and developers is to produce systems that deliver the services that users need in a form that users need them. If a system results in so much confusion then it's to those designers and developers that you should look to place the blame.
In fairness, simultaneous use of multiple IDs is rarely handled well by modern UI'd desktops and remote services. All systems assume you have one ID, and if you have the temerity to want more, then you must log out and log back in again, an idea unchanged since mainframes stalked the earth.
KDE has had a "Switch User" option on the Power/Session section of the menu for a long time. Opt for that and whatever's running stays running but you're presented with the login screen which you'd obviously need for the other ID* - there'd be something wrong if you could just waft over into a different ID without presenting any credentials. Log out of the other ID when you've finished and, again, you have to present credentials to get back to your original session but it's still there as you left it. I don't see anything wrong about having to provide ID & password, in fact I'd count it as a problem if you didn't.
* This is assuming you're not just wanting a terminal session for the other user in which case you can just su.
Yes. /home needs to sit where it won't be reformatted if you reinstall - which you might do, even for a version upgrade. Likewise if you have other a large data requirement. The other requirement which you might have that needs to avoid reformatting is locally installed S/W, /usr/local and/or /opt. But, yes, /usr on the root partition might not be a problem these days.
The other thing to remember is that in the Unix/Linux world we tend to have swap on its own partition.
Stuff has moved around a lot over the years. I can remember when everything was in /bin, /lib and so forth with /usr being for users' home directories plus a few things like /usr/spool. Then it got more and more crowded so /u was set up for users and that started to get crowded and it was split with a /u2 before /home was used. All these changes had a rationale, sometimes clearer than others. I still don't get the rationale for putting www and mysql data in /var. Yes I know it's supposed to be for changeable stuff but you're liable to find you have to reformat it to do a reinstall because, certainly with apt based systems, there's a install-related files in there and the installer seems to expect it to be clean. I tried not reformatting as an experiment and got errors so those files live in /srv now with links back to where they're expected.
"The upgrade notes also suggest running a utility called usrmerge, which simplifies the directory structure. This has been done since Linux 20."
This is something that's always been a bit variable across different Unix/Linus implementations. I'm a bit concerned about the logic of merging /sbin, though; IIRC the logic for this was that it contained stuff root might need in single user mode when /usr might not have been mounted.
It's items purchased up to but not including the 31st. Obviously.
I have come across a church register entry for February 30th. I'm not sure if it was a leap year or not.
Medieval courts almost invariably gave dates relative to saints' days or, between Shrove Tuesday and Trinity Sunday, the various moveable feasts. In a run of several years i found two calendar dates given. One was Friday April 1st. April 1st was a Saturday that year. I'm still not sure whether it was an April Fool's joke delayed by over 9 centuries or the scribe just wasn't used to handling actual calendar dates.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2025
