CJefferson

15,475 karmajoined vor 16 Jahren

Christopher Jefferson, St Andrews University, Scotland.

[email protected]

Submissions

Too much discussion of the XOR swap trick

152 pointsby CJeffersonvor 3 Monaten115 comments

comments

vor 4 Tagendiscuss

From the article there's this. If this isn't proof, well, I don't know what would be:

A 97-page internal presentation, “Teen Fundamentals,” in May 2020, described its goal as to “look … to biological factors that are relatively consistent across adolescent development and gain valuable unchanging insights to inform product strategy….”

That presentation conceded, “due to the immature brain they have a much harder time stopping even though they want to – our own product foundation research has shown teens are unhappy with the amount of time they spend on our app.”

vor 4 Tagendiscuss

Because most people can't set up a DNS blocker. There are lots of things we, as a society, decide we want to ban. I think social media is closer to illegal drugs.

We don't need to let multi-billion dollar companies maximise profits while ruining people's lives. We can just decide to ban them, as a society.

vor 5 Tagendiscuss

No, not at all. Students all have different voices. Yes there are similarities but if I read two pieces of work that sound in the same voice I assume there has been cheating — except this year everything is in the same voice.

vor 5 Tagendiscuss

This is my biggest concern. Speaking as someone who recently had to read 60 AI generated reports (the whole issue of how much students are using AI is one discussion), it was genuinely soul-destroying reading the same phrases, seem sentence structures, same arguments over and over. Depressed me the whole of the next day.

vor 10 Tagendiscuss

I feel this is on of those examples of where humanities and the softer sciences can teach us things. This is exactly how 'normal' languages work. New things are introduced and people (particularly older people) hate them. Over time common words (or even sounds) become shorter, in some cases disappearing entirely when they are 'understood from context'.

vor 10 Tagendiscuss

Has writing secure C really changed that much? Lots of the world still writes in C99, 27 years old at this point. I think they can use a 9 month old AI model is they want.

vor 10 Tagendiscuss

Wow, that might be the worst name for a project I’ve ever seen. I think every programmer who sees this is going to assume it’s a Python thing.

With regards the library itself —- I think it’s generally known the c++ standard library is a poorly designed mess in places but if you make an entirely new one you lose all the software already written, at which point why use C++ nowadays?

vor 11 Tagendiscuss

They are going to make it a thousands times worse.

It wasn't perfect before, but it at least took some time to fake a paper. The problem is now people can produce a very plausible looking completely fake paper in minutes. Peer review is in the process of completely collapsing, in fact I think it's already basically done.

The only way this might fix things is if we require all papers are completely reproducable (that doesn't help in subjects like biology of course. They can still provide all the experimental data in the rawest format possible which doesn't break any laws).

vor 11 Tagendiscuss

I think what upsets me the most isn't that companies do this, it's that no government seems to have any interest in stopping them.

To me, it should be a fairly easy, short, and extremely popular law that if you say you are 'selling' someone something, with a time-frame extremely clearly specified in a simple way, at least the same size as the price (say), you can't take it back later without giving a full refund. If they want to offer a multi-year rent, call it rent.

vor 12 Tagendiscuss

So do we get rid of all laws which distinguish children? Why bother with limiting selling alcohol to children if we can just let the parents do it?

My problem is I don’t think anyone is (seriously) suggesting we get rid of the laws protecting children in the physical world, and having nieces and nephews, they nowadays spend more time in the virtual world than the physical world, often with their friends so it’s hard to track what they do.

vor 15 Tagendiscuss

While you could send a letter anonymously, it was also trivial for the government to read any letter it wanted.

The problem is we have switched from a world where it was easy to let the government read selected communications (single phones, letters), to one where it's hard not to give them access to everything or nothing.

Personally, if there was some 'magic wand' way I could let the government keep it's previous levels of control in the internet age (they could individually pick users and put work into monitoring their communications, with a clear low limit on the number they can previously watch), I personally would.

But that's hard to do -- it's not obvious it's possible at all, so we need to define a 'new normal', but let's not pretend we aren't taking a huge amount of existing power away from governments with large scale encryption.

vor 17 Tagendiscuss

The problem is there used to be a fairly high correlation between ‘security report’ and ‘real vulnerability’. Not perfect but good enough. Now the two are almost entirely disconnected.

vor 18 Tagendiscuss

Many people run older versions of Mac OS X (because their machine won't upgrade), so their safari is aging -- they use an alternative browser, firefox and chrome support older mac os x.

Of course using unsupported OSes isn't the best idea, but Apple give you security updates longer than they give you 'fun' safari updates.

vor 19 Tagendiscuss

One big advantage I’ve found — people get attached to models (including me). With open models if you find one that works perfectly for you but the next version doesn’t, you can run the old one forever (or someone will for you)

vor 23 Tagendiscuss

In the UK I see a medical practitioner before any tests, always.

vor 23 Tagendiscuss

Even at a smaller scale, I was shocked to go to the hospital in China and literally the first thing done was a blood sample, scanned under about 30 metrics, took about 15 minutes after the same was take. The results were sent straight to my phone. That sample had some horrendous results, so I then skipped the queue and got straight to see a doctor who already had all my numbers up on screen before he spoke to the for the first time, meaning he could immediately put me on an antibiotic drip.

vor 28 Tagendiscuss

That's genuinely good to know.

The thing that often hits me nowadays is using pandoc (or pandoc via quarto) to go markdown -> latex -> pdf, which tends to bury these issues (and in markdown it's more tempting to just write emojis / unicode maths symbols.

Why use markdown, because I want good HTML output (which I realise is getting better in latex, but still not as good as markdown -> HTML.

vor 28 Tagendiscuss

I'd much prefer bad looking unicode characters to just skipping them and having broken output honestly. This is currently the main reason I'm considering moving from LaTeX to typst -- just always having broken output as I use more unicode (maths symbols, emojis) in my documents.

letzten Monatdiscuss

I tool which pulls all details from my Garmin and keeps a '5 day rolling average' series of stats -- my lifestyle means I can't keep up 'X steps a day', but I feel I can keep up 'average X steps a day for the last 5 days'. What's perfect is I can tweak it, I even added a special 'the weather is truely awful today, you can dip under average but you need to make it up later' option by getting it to automatically pull the weather forecast, or 'The weather is perfect today, really push'.

letzten Monatdiscuss

My advice -- give it a try. Chuck $5 into deepseek.com , and use this config (put it in a shell script, run ' . ./deepseek-claude.sh ', then just run claude as normal.

    export ANTHROPIC_BASE_URL=https://api.deepseek.com/anthropic
    export ANTHROPIC_AUTH_TOKEN= *** PUT YOUR DEEPSEEK KEY HERE ***
    export ANTHROPIC_MODEL=deepseek-v4-pro
    export ANTHROPIC_DEFAULT_OPUS_MODEL=deepseek-v4-pro
    export ANTHROPIC_DEFAULT_SONNET_MODEL=deepseek-v4-pro
    export ANTHROPIC_DEFAULT_HAIKU_MODEL=deepseek-v4-flash
    export CLAUDE_CODE_SUBAGENT_MODEL=deepseek-v4-flash
    export CLAUDE_CODE_EFFORT_LEVEL=max

I started by using it for some bigger reading jobs, particularly when I was near limit. Honestly, it's not quite as good, but it's much cheaper, and means I can carry on working. I also find sometimes it's good to ask claude and deepseek to consider code, how to polish, it see what they both say.