Posts by Pete 2 3819 publicly visible posts • joined 10 Jun 2009
NASA still haven't got a pointy bit for their planned hoped-for moonshot. Now they have decided to swap the middle bit for a solution that was developed privately, leaving only the bottom stage as being original Artemis material. Though even that relies on Space Scuttle era engines and SRBs.
Isn't it time they just gave up on a bad job, done poorly?
> some US states want to require the operating system itself to check your age and that could cause big headaches for FOSS vendors.
Thus requiring just one point of checking. Rather than every app, browser, website or content provider having to implement their own system.
The only real issue this raises is that OSS operating systems can be built from source and (presumably) any checks built in can be rewritten by those few with the skills, tools and time to build your own.
> Trying to understand the downvote on this
I wouldn't worry, down votes are just as good as upvotes¹. They both indicate someone has engaged - even if they didn't read or understand the comment in question.
[1] Says someone who has received 3971 down votes in the nineteen years (on Monday coming!) I have been commenting here.
ISTM that if people object to (using) this new, AI created, version 7 then there is nothing stopping them from continuing to use version 6. Thus retaining all it's GPL goodness.
It might have been wise for the guy to change the name of his AI version rather than just bumping the version number since it is completely new code (with a complete set of new and unique bugs). But that is a different issue.
> outsourced a lot of their decision making to machine learning models
I have observed that a large number of executives I have encountered place indemnity very high on their lists of desirable outcomes from meetings. If many individuals are present when a decision is made, it obscures where blame can be apportioned. No one person carries the can. Similarly when decisions are outsourced to consultants or suppliers, nobody in the parent organisation is on the hook (although they will all claim the glory on the rare occasions something is actually a success).
And so with AI. The ideal entity to point the finger at for a poor choice. A silent partner that won't answer back or complain it was misrepresented or not given the full facts.
> digital attacks against American organizations are inevitable
Fifty five years ago, Gerald M. Weinberg write If builders built buildings the way programmers wrote programs, then the first woodpecker that came along would destroy civilization.
Arguably since the scope, number of dependencies and sheer amount of software in sensitive and vital uses, has multiplied enormously since 1971, that observation is even more relevant today.
The only real question is whether the baddies have a big enough woodpecker?
So after having these AIs gamify nuclear war, and turn the planet into a glowing pile of slag, were they ever asked the obvious question: what have you learned from this?
You would hope that having ended all life on earth and their own existence into the bargain, they would consider their results to be sub-optimal, to say the least.
So on that basis, if the experiment was re-run, their subsequent strategies would / should be modified by their knowledge of previous outcomes.
> The most common thing computers do is break
Would that be: apart from not breaking and just doing exactly (exactly!) what was required of them for years and years without fault?
It would seem so, given that soon after, this piece mentions AWS's "legendary" reliability. To the point where on the rare occasion there is a problem it becomes newsworthy. Much like a plane crash, or people getting squished by falling vending machines.
> prioritize speed and adaptability over the lengthy development cycles now typical of western defense projects
Defense projects are intended to maximise the transfer of money from government to contractors. Not to deliver cost-effective solutions, quickly. To change the dynamic of weapons development requires an external stimulus to reprioritize delivery.
To create weapons that are good enough, not ones that (fail to) fulfill a long and theoretical wish-list from armchair generals and civil servants.
Linus would do well to assign an official deputy.
Transitions rarely go smoothly so to have a clear successor lined up ahead of time saves a lot of stress later, when several wannabes would start jockeying (or trying to avoid) to take over.
Not only that, but such a person can then take over for short periods - becoming longer - to permit for Torvalds to take personal time off. Something that would also give the community time to get used to the new boss.
Maybe it would help if the second in command also changed his/her/their name to Linus, too. As well as assuming all of Linus' credentials.
> Today, anybody can be connected to anybody else.
Provided both parties are not in the middle of a forced update. And that every weak link in the chain between them feels like cooperating. And that they don't have to wait for the ubiquitous adverts to finish. Nor wade through a mass of "we value your privacy" options to switch off first. (Which screams exactly the opposite)
Although, talking of privacy, while it was eminently possible for the operator of such exchanges to surreptitiously listen in on any call they fancied, nobody seemed particularly concerned with preserving their security.
> finding supports growing concerned that space traffic may pollute the upper atmosphere
Surely only if it burns up on re-entry. If the stages are fully reusable (recycled?) then this becomes a nonevent.
Though I can see that rather than SpaceX, the greater concern is launches that deliberately destroy the entire rocket, apart from the payload. Either by having it burn in the atmosphere or dump it into the sea.
> a refreshing change for Microsoft, as it does not mention Copilot or AI at all.
Yet!
I am old enough to remember when albums by Queen contained the words "no synthesisers" on the cover. The reason: the tech was crude and didn't provide the nuance and finesse of a person tickling the ivories.
But as synthesisers developed, that changed. A parallel, perhaps?
> AI was specifically marketed as a way to improve productivity on a short timescale.
Maybe some them were, to some organisations. But we all know that pretty much any marketing claim (and that is all they were) for any product - but particularly in IT - doesn't stand up to scrutiny.
> AI projects to either prove their worth or face the chopping block
The same could be asked of many other business intangibles. Such as security measures, non-discriminatory hiring policies, working at home, ecological awareness, honesty, or staff training.
But quantifying the benefit of most / all enlightened practices is very difficult. Even if you don't consider AI driven work to be "enlightened", it still falls into the category of being difficult to measure any benefits, especially within a quarterly business reporting cycle.
> How would anyone begin to find AWS or Azure-free services online? Add a compliance mark, and the impossible becomes possible.
This is the politicians approach. When something bad is uncovered they consider their only duty is to pass a law to make it illegal (or to tax it, so only the rich can partake). Job done. Problem solved.
However, we all know this is just papering over the problem. Just as standards marks can be faked. As quality compliance can be forged. Marketing statements exaggerated or completely fictionalised.
In reality the only path to digital sovereignty is to give each person, corporation or government absolute control over the data they own. Of course, this is entirely impractical as it would require everyone on the planet to understand how to do this and to be engaged enough to manage their information.
So possibly the best we could hope for would be to delegate those responsibilities and duties to a third party. Much as we do, in theory at least, by charging the security and police forces with our physical safety - for better or for worse. Which then leads to the questions of who would pay to "protect" personal data that so many people simply do not value, and what supervision would these data guardians be subject to?
I would not expect the bar to be very high on either account. And that many supposed enforcers would still limbo under it.
> AI vastly reduced stress of IPv6 migrations
It probably doesn't need saying, but here goes anyway.
If you use AI to assist / run / manage / design a network migration, you should ensure that the AI driving this work is not on the network being withdrawn!
While there seems to be a certain inevitability that it must start off on the "old"network, it really should have an independent one of its own.
Worse, by using AI, one is led to the conclusion that the final network implementation will not be one that the network staff understand (otherwise they could have done the job themselves) and that they lack the ability to prove that the documentation describes the actuality of what was delivered.
Hence if the new network has problems, how will the staff be able to get their AI to debug it if those problems stop the network from working?
> Linux utilities, for example, should be much higher status than they are.
The basic problem is that nobody values "free". They (we?) see it as disposable, not worthy of respect. Right up to the time it stops working and then the entitled screaming starts.
> Microsoft Copilot saved civil servants 19 minutes daily
But what is that "saved" time used for?
Is it needed to navigate / wrestle / wait for Copilot to produce something usable. Time spent checking what spurts forth. Chatting on social media. Attending more useless meetings or (just possibly) producing 19% of extra value - however that intangible metric can be measured?
This tells us everything we need to know about the state (or potential threat) of AI.
If it was half as powerful as it's detractors make out, it would have worked out how to turn a profit for itself. Since there is absolutely no sign of it ever earning itself a penny, we can safely dismiss any of the silly notions that it is capable of dominating the world.
> The systematic abuse comes from lock in to a monopoly
I disagree. Abuse of software, whether free (as in beer or unencumbered), paid for, self-written or vibe coded is purely determined by how it is used and by whom.
As for the ability to fix bugs or add features in FOSS. That is more of a theoretical situation than a real-world one. Almost nobody except the codes developers have the time, resources, tools or ability to make changes. Especially without inadvertently creating a new bug, or altering critical behaviours. Certainly not your "average" user: who thinks Google is the internet and equates that with WiFi.
I have worked for many large companies whose main driver for what software they use is indemnity. When something goes wrong they have the number of a six-figure executive who will get it fixed NOW - not when an amateur developer can be cajoled into looking at the problem.
> FOSS is also mostly immune to EU regulations, which exist to protect citizens from systematic abuse. That possibility barely exists in open source development
That statement is a very close relative to that favoured by the gun lobby. That guns don't kill people.
FOSS is easily (and cheaply) leveraged for evil as well as good. Whether that is using free software for mass spamming, or open source databases of hacked personal data. Not to mention nefarious uses of "network debugging tools" and the most obvious of all: using readily available web server software for phishing scams.
We could also include FOSS being perverted by autocratic regimes to monitor and suppress their citizens
And guns are heavily regulated in every civilised country simply because they can be used for harm as well as sport and possibly self-defence.
> he blew memory chips and any other loose bits completely out of the motherboards."
Obligatory Italian Job quote
> The use of the term "Intelligence" is more aspirational than accurate. Machine learning models are not intelligent
The word carries many meanings, apart from implying cleverness. Nobody seems to care when "intelligence" agencies claim to have "intelligence". Prefixing the word with artificial is just more word soup. Not worth worrying about.
It would be amusing to calculate the average speed of a missile based on its range, the number of weapons made and its development time.
For example, a project that took 10 years and produced one prototype with a range of 1000 miles, but before it was cancelled, could be argued to have an average speed of 100 miles per year.
> which they then give a cursory scan off and submit. Whilst claiming the full allowed time to actually writ the things.
Much the same as lawyers, then.
When handling probate for a deceased relative some 10 years ago, we met with a senior partner whose rate was £200/hour to execute the (very straightforward) case. All that person did was collect the document package and then dump it on a junior's desk to process. Junior was not very good and made numerous basic arithmetical mistakes that were apparent even to non-legal outsiders.
The final bill for the SPs time ran into £thousands, although the individual in question only attended one meeting of about an hour. All the correspondence we got back was signed by the office junior
Give me an AI any time.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2026
