Y
HN Search
Hacker News Search
new
|
comments
|
top
|
jobs
thom
searching PlanetScale…
1.
▲
2.
▲
3.
▲
4.
▲
5.
▲
6.
▲
19 ms
·
331.
▲
by
thom
2y ago
Things are tough all over. I’m based in Sheffield, and we have an Advanced Manufacturing Research Centre and obviously a rich history of manufacturing and engineering across the city that pioneered chrome steel. But the process for new ente
332.
▲
by
thom
2y ago
People who lament the economics of modern computing never got a £300 phone bill for a month of nothing but IRC and Usenet.
333.
▲
by
thom
2y ago
Really says something when manufacturing and space launch cycle times are faster than some software projects.
334.
▲
by
thom
2y ago
I agree these objects (just like colourised photos) help bridge the distance to the past. But I've had the same experience purely with text. If you read Cicero's letters and diaries, and then just imagine him - suffering writer&#x
335.
▲
by
thom
2y ago
Our kids play Roblox but we don’t let them chat or add friends. They just have FaceTime on in the background while playing with school friends. Seems fun to me, and none of them appear to have become demented.
336.
▲
by
thom
2y ago
I think whatever rules we tried to play, the nob jokes would persist.
337.
▲
by
thom
2y ago
I DM for early teens and I would kill for a D&D party that took it all too seriously. :)
338.
▲
by
thom
2y ago
Yep, my experience using ChatGPT as a helper is very good. It can very quickly cook up a custom monster, give it relevant stats, and create some flavourful actions and attacks to make it unique. Lately I've been trying to hook this up
339.
▲
by
thom
2y ago
You know you're old when you find yourself grumbling while using a decibel meter on your phone at 2pm in the morning, but I respect that they got away with it.
340.
▲
by
thom
2y ago
Still going strong. https://www.bbc.co.uk/news/uk-england-south-yorkshire-647769...
341.
▲
by
thom
2y ago
I agree, but this is a good thing! My default approach these days is functional tests for everything possible, and property based tests for anything particularly algorithmic or containing lots of edge cases, and no unit tests outside that.
342.
▲
by
thom
2y ago
The idea of parasocial relationships with infinite bandwidth seems sad, but not necessarily infeasible. I’ve no idea if current technology is there, but the idea of Leela streaming chess while trash talking opponents and answering questions
343.
▲
by
thom
2y ago
It does seem under resourced. Version 5 never appeared and I don’t know what 6 is supposed to look like. It’s still a great online resource, but having my own library, and access to the Hellenic and Roman Library in London, I’m happy to tak
344.
▲
by
thom
2y ago
Slides are cursed with multiple audiences and lifetimes. After an initial persuasive speech, perhaps some new initiative gains momentum, and the slides are often the first and only relic associated with The Project for quite some time. They
345.
▲
by
thom
2y ago
This is nice! I do use ChatGPT a fair amount as I’m still crap at Latin and Greek, but a lot of scholarship just plonks it in untranslated. It has always worked fine for getting the gist. I am extremely lucky in that while moving house my w
346.
▲
by
thom
2y ago
Are these parallelised? <3ms overhead for a clean database and test code to interact with it is very good!
347.
▲
by
thom
2y ago
This is what I do, it has an overhead of about 10-20ms per test and I’ve had zero flakiness. Absolute no brainier from my point of view.
348.
▲
by
thom
2y ago
I’m quite looking forward to a future where we’ve finally accepted that all this stuff is just part of the domain and shouldn’t be treated like an ugly stepchild, and we’ve merged OLTP and OLAP with great performance for both, and the wolf
349.
▲
by
thom
2y ago
It also has to be able to come and argue in the comments.
350.
▲
by
thom
2y ago
It’s not AGI when it can do 1000 math puzzles. It’s AGI when it can do 1000 math puzzles then come and clean my kitchen.
351.
▲
by
thom
2y ago
Ah, but can they do it on a cold, rainy night in Stoke?
352.
▲
by
thom
2y ago
While Ding did put himself under pressure in many games by taking long thinks in situations where it didn't really seem to benefit him, the pressure he put himself under in the final game was different. He forced a very uncomfortable e
353.
▲
by
thom
2y ago
I don't mind the classification that "machine learning equals AI plus time". To me, AI is any surprisingly effective use of machine learning. As time goes by we expect more, but you'd have to be the most cynical curmudge
354.
▲
by
thom
2y ago
What if it's a big hoax and we create a better world for nothing?
355.
▲
by
thom
2y ago
I’m not paranoid about Google, but smart TVs inevitably come with dreadfully slow boot times, clunky animated menus, and just generally lots of crap to work around, even if you primarily interact with them via HDMI. I miss TVs that respond
356.
▲
by
thom
2y ago
I’m not really married to that form factor. I guess I’m more curious about HDMI splitters or pass through and what people do there.
357.
▲
by
thom
2y ago
What’s my best bet if I just want a normal computer monitor hooked up to some sort of sound bar, and I’m happy using something like a Chromecast for all TV needs? Or are there monitors with good enough speakers now?
358.
▲
by
thom
2y ago
What features are you seeking that aren’t available? There are many powerful, heavy laptops. There are many 16+ inch form factors. My current laptop has dual screens! I agree it can be frustrating when the mainstream focuses on things you d
359.
▲
by
thom
2y ago
For anyone that wants to test their mettle, the FEN of the key position in this game was: B7/8/4b3/4kp2/5Rp1/6P1/1r6/6K1 w - - 16 55 Give yourself 10 minutes and 30 seconds increment as White and see if yo
360.
▲
by
thom
2y ago
I don't think 26. a4 was fighting like a true warrior, it was more giving up half your kingdom in the hope that your opponent will then accept a peace treaty.
More ›