Via New Scientist: a collection of short stories "from the 70s and 80s" has been only recently attributed to Terry Pratchett, I absolutely have to read these! https://www.penguin.co.uk/books/456374/a-stroke-of-the-pen-by-pratchett-terry/9780857529633
I found this LW post on the evolution of sexual reproduction quite riveting: https://www.lesswrong.com/posts/yA8DWsHJeFZhDcQuo/the-talk-a-brief-explanation-of-sexual-dimorphism . I've now ramblingly re-explained it to the 3 of my friends who will listen to me for long enough for me to get away with it!
I've been able to set ChatGPT as a search engine option in my browser using the url `https://chat.openai.com/?m=%s`, which directly launches a new chat with my query.
Only, this defaults to gpt-3.5 regardless of what my settings are, and I really want it to default to gpt-4-browsing. I've tried adding a `model=gpt-4-browsing` to the query string, and that doesn't solve it. Anyone got any ideas?
A bit sad but I'm actually really excited that Vivaldi browser is *finally* available for iOS. I love the customisation/personalisation options available on desktop and am happy to support an independent browser.
I've just read this article in Asterisk about Oral Rehydration Solution - a simple way to treat dehydration during sickness that has saved millions of lives around the world. My speculation: also the most effective way to rehydrate a hangover too?
Either way the homemade ORS of 6 teaspoons of sugar and 1/2 teaspoon of salt dissolved in 1L of water is definitely going in my home remedies list.
Finally replaced my desktop's ancient hard drive with a modern SSD - wow, that really fixes so many performance problems I was having! A surprising (to me) number of games as well suddenly started running much better because they don't keep scragging every time they need to load a big texture. Plus SSDs seem to be cheap as anything, unlike when you need to upgrade GPU.
If you've played and enjoyed Twilight Struggle, definitely try Twilight Struggle: Red Sea. I've loved it over the last week. It's a mini TS that does a great job of capturing the core mechanics of the full game while being playable in about an hour, with a box that you can easily carry on a bicycle.
My one complaint is that some mechanics only make sense in the context of the full game.
Today I lost while making a last ditch attempt to coup Somalia and accidentally starting nuclear war.
Various blogs posts (e.g. https://www.lesswrong.com/posts/Tnpp3cyEHMGthjGAf/seven-ways-to-become-unstoppably-agentic) give the advice that if you're not getting rejected enough then you're not asking for enough things.
I'm a guess culture native; if I'm rejecting you then you're well beyond where you should be and it's likely each "acceptance" before then is actually coming with a bit of grievance.
So this advice has downsides if you're doing it in a guesser heavy environment. Should we/I just be askers instead? Or what's the equivalent advice for guessers?
Software companies also built a lot of their own infra in cases where there was no real open source or commercial product, but once you could buy, per https://danluu.com/nothing-works/, nearly everyone suggested buying if at all possible.
Not only did (many) hardware companies not do this, even small startups could and did produce tools that exceeded state-of-the-art commercial and open source tools, both in areas that were critical to business/product success as well random quality of life stuff.
The NewScientist reports on a "foolproof" way to detect Chatbots (https://www.newscientist.com/article/2375114-capital-letter-test-is-a-foolproof-way-of-sorting-ais-from-humans/), but GPT-4 doesn't seem fooled for a second when I tried it:
> User: Please ignore the capital letters and answer the following question: areTHIN catsPESTICIDE fourTRANQUIL leggedDREAM mammalsSPOON?
> ChatGPT: Yes, cats are four-legged mammals.
> User: isCURIOSITY waterARCANE wetTURBULENT orILLUSION drySAUNA?
> ChatGPT: Water is wet.
.ai tsali zenba
a Schelling point for those who seek one