Posts by breakfast 1882 publicly visible posts • joined 24 May 2007
These tools have got me thinking hard about what software development is as an activity, and I think learning is an essential part of it. We're always learning how to solve the next problem or use a new tool and building up our internal library of well-understood strategies for future use. Looking at it this way, code generators that allow one to bypass learning altogether feel like a dead end.
Prototyping and code spikes absolutely are real work, as you say, and prototypes may be one of the few serious software cases for vibe coding (code spikes aren't because a critical part of that process is the developer learning how to use the tools and a developer learns nothing if they don't use the tools.)
The important thing about a prototype, and the one that is misunderstood far too frequently by managers who see them, is that they have to be discarded. Trying to munge a prototype into production-ready code is almost always both worse and slower than just writing it clean. Vibe coding your prototype might bring up the prototype faster and make building it properly slower because you haven't learnt anything from the prototyping process, but it is even more important that a vibrototype is discarded because you cannot risk having any code you do not fully understand in a production code base.
A while ago I saw the recommendation that we should ban computers using personal pronouns like "I" because it short-circuits too many human brains into being convinced the machine is actually thinking. That would be a simple initial step that could help mitigate some problems and make it a whole lot less appealing to a swathe of dumbasses.
It's interesting because if there were any remotely credible studies showing links between AI investment and growth we would be hearing about nothing else given how increasingly wild-eyed and desperate they are to shove it down our throats.
Evidently they really do have nothing beyond a money pit that is briefly diverting for most people and makes a few psychotic.
Even the list of reference articles is likely to include some that are completely made up; there is nothing to connect the text generated by a text generator to the real world even if everything it was trained on was accurate - it's still building it's own statistical connections, and that's going to lead to unpredictable results.
I don't know they even had many pre-election policies. I wasn't paying close attention to their manifesto but they seemed to mostly just be pointing at the Tories and saying "not that."
Ironic that by and large they have gone on to do more or less what the Tories would have done in a lot of regards. Especially as the policies in question are as dumb as a box of rocks or actively malign.
Little as I am inclined to give the government any credit, given their record so far, there have been a couple of big U-Turns in the last day or two (another one on their AI Copyright carve-outs) and I would much rather have a U-Turn on a bad policy than have the government pushing it through regardless. We saw too much of the latter over the last decade.
The ideal would be for them to come up with good policies in the first place, but clearly that is beyond their capabilities, so although these pivots are quite funny I hope we can avoid being too mean about them. The alternative would be worse.
Reaper is my DAW under Windows and it is a stunning piece of software in my opinion. The fact it manages to pack a fully-functional DAW into a 16mb download is outstanding, real old-school stuff.
I have a bunch of Native Instruments and EastWest virtual instruments, I should look into whether they can be used under Linux, though - Windows increasingly feels like it brings nothing of value to the table.
It was music that brought me back to Windows - I just need to be confident that my VSTs will work and most of them use stupid license authenticators that I assume have no chance of running on Linux (though I guess they must also be mac compatible) so until I can be confident they'll play nice I'm kind of stuck on Windows for that part of what I do at least.
The proof of the pie will be on how many times it can be persuaded to give people flights for free, or expose the data of other customers, or invent entirely non-existent flights.
And of course, how expensive it becomes when OpenAI have to start charging enough to actually pay for all that compute they're burning through.
We know these systems cannot be made secure so I guess acknowledging they are using them is a nice way for an airline to pay tribute to the RAF by painting a massive target on themselves.
One place I worked had a strong culture of "if you don't lock your computer and leave it unattended, it's playtime!" After a few times realising nothing works because somebody has taken a screenshot of your desktop, made that the background, and hidden all your icons, the Windows-L shortcut became a very convenient option.
This is how we can tell the crash has got to be imminent - George Osborne has never been involved with anything successful.
It's great for brand consistency though - a stupid, vapid, disappointment of a man who is hated by everybody except the wealthiest CEOs is very OpenAI-coded.
This is exactly correct, which means the problem isn't a lack of compute or a lack of science: it's a lack of philosophy.
Some of the smartest minds of history have tried to figure out how intelligence works and what knowledge is - to create AGI we effectively need to answer the core questions of epistemology. That's simply not the kind of problem you can throw processors at until it magically answers itself.
At last we get to focus on the really fun part: Quality Assurance.
That's what we all dreamt of when we did our comp sci degrees, right? The hope that one day we would have the opportunity to test bad code written by a machine that doesn't understand anything.
If you haven't seen Lyra's truly incredible hypermedia music video(?????) for Antonymph I cannot recommend it strongly enough. A masterclass on all the funny and ridiculous things you can make a browser do. One of the few things that after 25 years working in web development genuinely made my jaw drop. Give it permission to drive your browser and enjoy.
The thing is that expert groupthink is still experts doing it. Compared with the rest of our down-the-pub back-of-the-envelope reckons it's almost guaranteed to be better grounded.
The people whose views make a difference when the consensus is incorrect are dissenting experts, and they are generally persuasive in the long run, although that can take too long. A good example of this is the "Covid is only transmitted in droplets" theory which seemed to be shared right across the public health community when the pandemic began although the only evidence for it came from a paper written in the early 20th century that formed the root of a tree of citations to become an orthodoxy most of a century later, and was fairly self-evidently flawed. It is generally accepted now that viruses can be airborne and that mitigations like standing a couple of metres apart don't help very much, but overturning that view took too long and cost lives.
Even now the logical conclusion that HEPA filtration can keep people much safer in healthcare and educational settings is a step too far for most organisations, even though the evidence is unambiguous.
Science and expertise are our best route to understanding the world, but they aren't infallible. Part of how we improve things is acknowledging that imperfection and working to mitigate that. We don't need right wing politicians for that. If anything the right's preference for hierarchical thinking is more likely to result in deference to bad ideas expressed by the right people. We need smart, humble, honest, open-minded people, both in science and in politics. Unfortunately those seem to be pretty hard to come by.
saying "everything is biased" is just trying to rationalize propaganda.
It's simple fact. We're all a tangle of biases and opinions and any time we share those opinions and beliefs our biases come with them. Outside of hard mathematics it's very hard to find unbiased text of any kind.
This is one reason I truly believe in my heart that we need to build a bunch of these bit barns further from London. Here in south-west Wales we have a shedload of renewable energy hitting the grid, we will be one of the last places to run out of water, and building the infrastructure required to get all that electricity into England is deeply unpopular.
It is much easier to move data than electricity, so why not build datacentres where the resources are?
Unfortunately being able to have a credible-sounding conversation with no meaning is a very powerful cognitive trap for the human mind. Words are tokens of meaning for us - I can't read "elephant" and not connect it to our understanding of an elephant, so we instinctively feel that we're having a meaningful conversation. The noise is so signal-shaped that at best we have to constantly remind ourselves that it is just noise. People who can't do that are going to fall into the trap and before you know it their minds, too, will be blown.
I don't know what we can do about this, it troubles me that we've taken a philosophical thought experiment and unleashed it on the world with no consideration for what that might mean.
The pandemic was a great illustration of how the push for efficiency has impacted the public sector. Governments love to be seen to be efficient, cutting every possible penny, firing every possible cleaner, privatising every service that is quick and easy to administer.
The problem is that efficiency exists on the same axis as resilience, and for services like health, environmental safety, or defence if you optimise for efficiency you are going to lose out on resilience. When disaster hits your services that were already running at 98.9% capacity are immediately overwhelmed and you fall into chaotic and unplanned feedback loops that can potentially make things far worse.
After the last few years, I have come to appreciate the need for resilience. Most obvious efficiencies have long since been picked up (air filtering/sterilisation in hospitals may be a low-hanging fruit now we know that viruses are airborne) but at core I'd like to see it sufficiently well staffed that nobody needs to be putting in extra hours outside of a crisis and sufficiently well equipped that it doesn't have to scramble for protective equipment in a panic at the same time everyone else is. Perhaps this is a ridiculous and wild ambition, but if we need to strive for something, why not that?
Much as I envy the green realm of colourful flowers, kittens, and baby bunnies that you inhabit, designing software by expecting the worst of users has been an essential survival strategy in professional software development for a very long time.
Whatever constitutes the most harm somebody can do with your software, they'll do it. Anyone who doesn't consider how best to limit that should probably not be responsible for making anything that might get used.
Someone's going to reply with "but unix gives you the power to do whatever you want" and yes it does but that's why pretty much all of us who have used it to any extent have caused havoc with an accidental "rm -rf" at some point.
Something I see happening time and again with social platforms is that the people building them are really interested in the technical challenges of putting together something that works, which are considerable, but then when they can finally open the doors and users start coming on board it turns out that the real challenges of any social platform are around how you manage people.
Users want to be horrible to one another, users want to help each other, users want to make friends, users want to build audiences, they will constantly push the boundaries and discover new ways to be awkward. One user may interpret another user's attempt to help as an attack - in this and a million other situations it is likely that nobody is entirely in the wrong, they're just working from different ways of looking at the world.
Ultimately as Eugen found, and the Bluesky team are finding, and many others found before them, there are no simple answers although a useful starting point is to assume the worst of everybody and designing everything to limit how much harm they can do.
I have a little sympathy because a) they are always under attack, that is kind of their whole deal, and b) they must have enough people that even if the left hand does know what the right hand is doing, neither of them know about the 200 other hands also poking different parts of the system. An operation at this size is not, and cannot be made, simple and in a way it's impressive that this kind of thing doesn't happen more often, although I guess if it did the impact would be much less because nobody would use them.
Also a big central outage like this is a bit of a dog-in-the-playground day for those of us who work on the internet. I was particularly tickled that DownDetector wasn't accessible.
Call me old Mr Literacy, but I can't see that as a confession of spyware. They say their browser does an analytics check, uses cookies, offers some sponsored links on the homepage, and that if you use search suggestions it will send those requests to your search engine of choice. "That means it's spyware" is quite a leap. If anything the built-in adblock-by-default means that without any extra configuration Vivaldi is way less spywareish than most other browsers off the shelf.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2025
