Oh, how insensitive from my part. I didn't realize this was done to memorialize Sir Tony Hoare. Thanks for the answer, and apologies for such a frivolous question.
And how do you actually identify who should pay that $0.000713? And who should receive it? How do you make the process effortless, so the user doesn't have to spend 5 minutes registering on a website, just to send $0.000713?
Now make it work 10,000 times per day, for every page you visit, posts, news, short form content you scroll, long form video you watch. And multiply this by billions of users.
And once you've done that, how do you deal with spam, bots? How do you prevent invalid traffic? Fraudulent chargebacks? And how do you take quality into consideration (NYT prob want to charge more than my crappy personal blog)?
Transferring money is one small element of large and complex equation.
Advertising is not perfect, but it's the best alternative for a free and open web I have seen in my 30+ years online. Subscription works for large ticket items (and for the affluent minority), but it doesn't solve the other 95% of cases.
Coincidentally I was just listening to your interview with The Pragmatic Engineer [1] this morning. Loved hearing the stories of early days at HashiCorp, taking it public, and the near-miss with VMware.
It also got me wondering how things would be different if you haven't crossed paths with the guy who unplugged your mouse :) It's fascinating how life is full of these small yet defining moments. We don't always appreciate them right away, but beautiful to look back.
Thanks for Ghostty! It has been my daily terminal driver for the past year.
I was also just recommended this interview on youtube. honestly it makes sense if the algo decided it was the right time to recommend this video and resultantly this post is making it's way to front page of HN
I subscribe to the channel, so I had already downloaded the episode. But glad PE is getting some algorithm-love; it's a great channel/podcast. Gergely is a great interviewer.
If you read the transformer paper, or get any book on NLP, you will see that this is not magic incantation; it's purely the attention mechanism at work. Or you can just ask Gemini or Claude why these prompts work.
But I get the impression from your comment that you have a fixed idea, and you're not really interested in understanding how or why it works.
If you think like a hammer, everything will look like a nail.
I know why it works, to varying and unmeasurable degrees of success. Just like if I poke a bull with a sharp stick, I know it's gonna get it's attention. It might choose to run away from me in one of any number of directions, or it might decide to turn around and gore me to death. I can't answer that question with any certainty then you can.
The system is inherently non-deterministic. Just because you can guide it a bit, doesn't mean you can predict outcomes.
The system isn't randomly non-deterministic; it is statistically probabilistic.
The next-token prediction and the attention mechanism is actually a rigorous deterministic mathematical process. The variation in output comes from how we sample from that curve, and the temperature used to calibrate the model. Because the underlying probabilities are mathematically calculated, the system's behavior remains highly predictable within statistical bounds.
Yes, it's a departure from the fully deterministic systems we're used to. But that's not different than the many real world systems: weather, biology, robotics, quantum mechanics. Even the computer you're reading this right now is full of probabilistic processes, abstracted away through sigmoid-like functions that push the extremes to 0s and 1s.
A lot of words to say that for all intents and purposes... it's nondeterministic.
> Yes, it's a departure from the fully deterministic systems we're used to.
A system either produces the same output given the same input[1], or doesn't.
LLMs are nondeterministic by design. Sure, you can configure them with a zero temperature, a static seed, and so on, but they're of no use to anyone in that configuration. The nondeterminism is what gives them the illusion of "creativity", and other useful properties.
Classical computers, compilers, and programming languages are deterministic by design, even if they do contain complex logic that may affect their output in unpredictable ways. There's a world of difference.
[1]: Barring misbehavior due to malfunction, corruption or freak events of nature (cosmic rays, etc.).
But we can predict the outcomes, though. That's what we're saying, and it's true. Maybe not 100% of the time, but maybe it helps a significant amount of the time and that's what matters.
Is it engineering? Maybe not. But neither is knowing how to talk to junior developers so they're productive and don't feel bad. The engineering is at other levels.
You can reload the page. It's an animated gif that shows the full app functionality.
But your feedback is still valid. It's unfortunate that OP used the login screen as the last frame; it'd be better to show the app as the last frame, or just loop the animation.
Dr Dobbs was pretty good until almost the end, no? If memory serves me well, I recall the magazine got thinner and more sparse towards the end, but still high signal-to-noise ratio. Quite the opposite of Ars T.
Huge debt of gratitude to DDJ. I remember taking the bus to the capital every month just to buy the magazine on the newsstand.
I would go to the library on my bicycle to scour for a new copy of DDJ as a 10 year old.
I had dreams of someday meeting “Dr.
Dobbs.” Of course, that was back in the day when Microsoft mailed me a free Windows SDK with printed manuals when I sent them a letter asking them how to write Windows programs, complete with a note from somebody important (maybe Ballmer) wishing me luck programming for Windows. Wish I’d kept it.
It's a common trope. (Some) artists will often convey the same message; art should be judged on how hard it was to create. Hence why some artist despise abstract art or anything "simplistic".
We forget that human consumption doesn't increase with manufacturing complexity (it can be correlated, but not cause and effect). At the end of day, it's about human connection, which is dependent on emotion, usefulness, and availability.
I mean, I CAN see the value in pushing the context summary to git. We already have git blame to answer "who", but there is no git interrogate to answer the "why". This is clearly an attempt to make that a verb git can keep track of. It's a valuable idea.
I also seen examples of it before. I've got opencode running right now and it has a share session feature. That whole idea is just a spinoff on the concept of the same parent that led to this one.
In a way making the cards helps a ton to learn the content and decide what's really important to retain. On the other hand, it's such a slog that I usually end up relying on community cards, or skipping it altogether. The MCP idea may be a nice middle ground. Will give it a try for an upcoming exam.
Ha, same! On a TRS-80 Color, nonetheless. But I think I used four times, because no one else in the country had a BBS at the time (small city in Latin America).
It took a couple of years until it would catch on, and by then 1200 and 2400 bps were already the norm - thankfully!
reply