brainwipe

Met him at GenCon UK in the 1990s. His Doc-From-Back-To-The-Future energy was absolutely mesmeric. He had five towers of D10s in a plastic tubes. Three towers was the standard dice we all had in our bags and two were towers of his Game Science d10s. The standard dice towers all had different heights but the two game science towers were identical heights. He explained that for the regular dice, the smoothing process (which made them nice to hold!) changed the shape enough that they weren't regular sized anymore. The Game Science dice weren't smooth (sharp edged) and so they kept their shape. Of course, if you want a die to roll fairly, it needs to be a regular shape.

I bought a Zocchihedron cos we were playing a game that used percentages (Icar RPG) but rolling it was hilariously haphazard as it was essentially a ball! Loved it tho. I later received the game science dice set as a gift, which I still have. Sadly little time to play these days.

I code modern C# during the day and Unity for fun and I found this a really good roundup.

It's also worth noting that Unity does all sorts of OO-breaking filth before passing to IL2CPP.

XML was much better than what was there before - which was a different standard for every endpoint and often no structure at all.

XML allowed you to use tools to build automatically. We have other better tools now but back then it was like magic. Download an XSD (the more common option to Relax NG but not superior IMO), point a pre-built tool for it and it'd build strongly typed model classes and validation checkers. Then, when you called the service, chances are it would work first time. It could also be used to write the specification too. That was unheard of before. Often the spec you'd get for service was different to what the endpoint served because keeping documentation up to date was not a priority.

XML then got a little overplayed. For example, XSL transforms allowed you to turn one XML into another and because XHTML existed you have people building entire front ends in it (not recommended). You ended up in a weird hinterland where XML wasn't just for representing structured data but it had built in functionality too. It was not the right tool for that job!

I've not needed it in a long time as I prefer lighter weight formats and I don't miss it.

Just my take, others will have their own!

I own my grandfather's slide rule, he was a master toolmaker for Rolls Royce (aerospace engines) in North London during WW2.

The title is irritating, conflating AI with LLMs. LLMs are a subset of AI. I expect future systems will be mobs of expert AI agents rather than relying on LLMs to do everything. An LLM will likely be in the mix for at least the natural language processing but I wouldn't bet the farm on them alone.

Two things! - Videogame, you live inside a mechanical ladybird called a Clomper, which you control by making pipes to power machines with steam. https://store.steampowered.com/app/2349380/Clomper/

- Building plastic self organising maps (Lang 2002) using Python CUDA build to parallelise the more expensive bits. Also fancy building the directed graph half in Unity 3D.

- Also doing some data engineering pre-training and AIIA at work but no deets, sadly.

I was asked by an SME to code on a whiteboard for an interview (in 2005? I think?). I asked if I could have a computer, they said no. I asked if I would be using a whiteboard during my day-to-day. They said no. I asked why they used whiteboards, they said they were mimicking Google's best practice. That discussion went on for a good few minutes and by the end of it I was teetering on leaving because the fit wasn't good.

I agreed to do it as long as they understood that I felt it was a terrible way of assessing someone's ability to code. I was allowed to use any programming language because they knew them all (allegedly).

The solution was a pretty obvious bit-shift. So I wrote memory registers up on the board and did it in Motorola 68000 Assembler (because I had been doing a lot of it around that time), halfway through they stopped me and I said I'd be happy to do it again if they gave me a computer.

The offered me the job. I went elsewhere.

I've accidentally been using an AI-proof hiring technique for about 20 years: ask a junior developer to bring code with them and ask them to explain it verbally. You can then talk about what they would change, how they would change it, what they would do differently, if they've used patterns (on purpose or by accident) what the benefits/drawbacks are etc. If they're a senior dev, we give them - on the day - a small but humorously-nasty chunk of code and ask them to reason through it live.

Works really well and it mimics the what we find is the most important bit about coding.

I don't mind if they use AI to shortcut the boring stuff in the day-to-day, as long as they can think critically about the result.

It's scrubbed for today: Blue Origin on Twitter: https://x.com/blueorigin/status/1878715911563313651

Wonderful! My PhD was in stream anomaly detection using dynamic neural networks in 2003. Can't wait to go deep through this paper and find out what the latest thinking is. Thanks, OP.

This has been my life for the last 10 years and I know who my children really are. They're evil and I love it.

The only one that directly annoys me is not being able to have threaded comments at the PR level. https://github.com/github/roadmap/issues/552 You can do it with "quoting", which is fine if there are two of you but turns into a mess if there's more than that.

They've said that they're watching the discussions for feedback, so I hope they listen and implement that one.

Happy that they are being transparent (rather than letting the issues rot), annoyed that they appear to be prioritising marginally useful AI stuff for basic UX.

Then I would ask anyone to keep building on knowledge rather than training for an arbitrary benchmark. Filling your short-term memory with knowledge to be then dumped straight after will get you a good exam mark but doesn't mean you have anything close to a solid grasp. Most schools (in the UK in particular) optimise for grade outcome because that's how they are judged, that's not the same thing as being good at a task.

Exams are a poor measure of proficiency. Proficiency is gained by doing and stretching a skill over time. You can measure that in small increments than a periodic exam. At the end of a period, a student would have a body of work to demonstrate proficiency rather than relying on a single day.

When I taught at university there was a disparity between exam grades and the physical body of work they had submitted over the years. You'd see the grade and be shocked the did so badly. The grade reflected proficiency in examinations, not in the subject.

IMO education is still built around Victorian structures and needs to be reworked from examinations downwards. Examinations are an exercise in being good at examinations, not proficiency in the subject. Once you strip that away the you wind back all the structures that feed it. You can see this working at schools designed for the neuro diverse. Those students simply can't sit and listen to a teacher all day, so each student learns in their own way and are better of for it.

Arguing about the effectiveness of edtech is like complaining there wasn't a viola on the Titanic's band.

Not the case universally. My son's school in south-central UK does everything via Google Docs or apps specially built for distance learning. If it weren't for Minecraft mods, he'd have no concept of the file system.

Articles like this remind me how little the browser environment has moved on in recent times. Now all the work is done by the frameworks.

Independent researcher without academic address; can't get in. Best of luck.

I give it 6 months before someone goes to trial for sharing AI-generated CP and is acquitted.

ESET started scanning SSL certificate chains for nodejs, which is a good feature. If you're using node without a full chain then it will block the requests. Unfortunately, they did so without warning, which broke our GitHub Actions Runner and our dev machines.

ESET deployed the change silently, breaking nodejs implementations and leaving people scrabbling around trying to add node to the SSL scan exception list.

Is it reasonable to expect some warning before a change like this?

I might be forced to choose a replacement. What's your goto brand of antivirus/botnet detection for server and workstation with centralised cloud management?

All of the above! Additionally... I think AI companies are trying to steer the conversation about safety so that when regulations do come in (and they will) that the legal culpability is with the user of the model, not the trainer of it. The business model doesn't work if you're liable for harm caused by your training process - especially if the harm is already covered by existing laws.

One example of that would be if your model was being used to spot criminals in video footage and it turns out that the bias of the model picks one socioeconomic group over another. Most western nations have laws protecting the public against that kind of abuse (albeit they're not applied fairly) and the fines are pretty steep.

Thanks for the tips! I'll admit not having done enough research yet - you've helped me do some shortcutting.

Article is 3.5 years old, best put the year in the title, OP!

I think the definition of serverless is too narrow (AWS lambdas/ Azure functions) and that serverless is really "I want to build apps, not manage infrastructure". That's not the same thing as putting everything onto functions.

I have a monolith we're preparing to move onto a managed container system (prolly AWS AppRunner). I don't want to manage K8 if I can help it, our app doesn't need complex server architecture.

Personally, I think it's just the next layer of abstraction up. Some won't benefit from serverless in the same way some are better off with in-house tin. I know some that need custom chipsets in-house, so can't even buy a stock rack server! However, many many web apps don't really need the control and will probably use facets of serverless over time. But it is still a revolution for old people like me that just don't want to manage servers anymore!

Having XSLT flashbacks. Send help.

Additionally - the BBC would put "how to code" programmes on the TV. That's how my neighbour got started when I was knee high.

Trying to program the web front end like you do the API backend feels like an antipattern. I'm all for productivity but you eventually hit issues that cannot be resolved because the client/serverness of your solution has been abstracted away from you. Those of us with WebForms scars remember those days.

Like the article suggests right at the end, I want a C#-compile-to-wasm with new language structures for common web browser features such as shadow DOM. Perhaps also without the .NET core lump unless you really need it - and even then importing only the dependencies you need. I almost love Typescript but only because I spend my life wishing it was really C#.

Blazor looks cool but it's not quite native enough to client/server. I've been burnt by Silverlight and have a lots of ReactJS at work so the benefit isn't quite worth the cost and risk. I wonder if in a future role I might be tasked with a greenfield app for which it's a brilliant fit but I can't see that in any of the SME roles I've had so far.

The piece of paper is the nub. You really need that to convert it to some other ID, cos no-one really checks.

Ideally, it needs to be signed off by an institution that your local passport office accepts. For my latest UK passport, I had to send off the real PhD certificate and it came back within 10 days. My friend (also UK citizen) got his PhD from University of Warsaw and it took months to verify. I am not sure what that process was but the issuer of the PhD is important if you want it legit.

Only my first ever job out of uni checked that I did a PhD there, nowhere else has.

In the UK, the passport appears to be the root ID, so all other renewals check that (driving license is the only other one I've done recently).

I think @edent's answer is best: PhD by publication is best - then bind them together with intro (which can be a survey paper), conclusion and any additional references. I know a Comp Sci PhD did this recently (2016?) at UK uni. Cracking PhD it was too.

For journal publish, yes but not for doing the actual work. If you find something interesting, you can go fast.

"At this stage, it’s still unsure that it will achieve such levels where no human is necessary"

Wishful thinking. You might be unsure but the copywriters struggling to find work might disagree.

You're quite right that the recipe is abused by a cook. I like that analogy!

However, I think a vanishingly small number of corporations care enough about the recipe to cook it properly. The recipe itself can be abused. A better recipe would be safe from abuse.

I'm not sure what that recipe looks like.