61 karma · joined December 24, 2021
wegfawefgawefg - I bookmarked this and worked through it more carefully when I had time, I appreciated the learnings.
Me: “staircase wit” is a very charming phrase. What are some other unexpected phrases that might also be fun to know?
ChatGPT: <I originally included the response, but you can ask it yourself if you care to. I feel this was a spur-of-the-moment idea and, rereading it, I feel it’s just like Googling something and posting the result. Not really a high quality contribution on my part.>
I now work hard on meditation-style “make your mind blank” effort at sleep time. Any worry I have, it can be handled tomorrow. (But I have to in good faith have a plan to address it in the morning.) This has made it much easier to fall asleep.
1. Ripeness-window is days, maybe even a week, and they can be super squishy and still not turn brown.
2. Because of #1, buy a good number of avocados when you notice you’re out and immediately toss them in the fridge. Maybe work to find that one in the store that is ready to go for today/tomorrow. Sure for the next few days you might not have an avocado ready, but…you probably don’t have to eat avocado every day so you’ll be ok.
3. If you really need avocados and you’re annoyed at the unripe avocados in your fridge, take one out and put it on the counter for tomorrow.
So much less stress to immediately use that purchased avocado. Net-net so much less wasted avocado.
From what I understand, PIPs and documentation save money compared to going through a lawsuit. In the event of a lawsuit, the “at will” employment principle would mean that the company would almost certainly win the suit (assuming not-illegal-discrimination), but it would be expensive to go through with the suit. Comparatively, a PIP, lots of documentation, and keeping the employee around for 6 months longer is a cheaper alternative.
TBH, if AI makes a lot of legal action cheaper, this calculus might change. There’s a business product here: an AI mediator that all parties agree to use at the start of employment. Looks at all documentation at time of firing and says: “based on this, employee would average $X out of a civil suit. Company would average $Y to defend. Cut employee a check for func($X, $Y) and by cashing it employee agrees to relinquish rights to further legal action.”
The bands appear to be stacked older-format-lower.
It’s visually interesting, but without a vertical scale I’m immediately suspicious.
Spend 8 hours slowly digging an access tunnel on Mars, 8 hours in the fully immersive fantasy world of Unicornia as a level 85 troll-mage, 8 hours sleeping. Rinse, repeat.
Without the mental escape, there would be sadism and brutality.
My working theory is that progress was really slow until the written word, incremental improvements but still slow until the printing press, and pretty quick thereafter (until computers, and then the wheels came off). But…I would love to read a history from someone who actually studied it, and said “look at the pace of discovery from 1000BC-200AD, 200AD-1200AD, 1200-1600, etc.”
Plus, (a) reviews for the electric VW aren’t impressed, (b) [charging] network matters, (c) pay the VW “premium” for first-gen tech? Seems like a poor bet.
Thanks for writing this, it’s been bothering me for a year, and without your comment I wouldn’t have thought about the possibility of a connection.
Part of it seems to be that the UK based companies don’t expect massive success. They have a thing, it makes some people at the top a pretty good living, that’s enough. They obviously could create massively successful companies based on the talent available, but no one wants to.
Or possibly more accurately: the people who want to realize that proximity to other like-minded people is critical to success, and they leave.
Every originally-from-the-UK tech worker I know in the US is well aware of how lucky they were to get out.
This book is probably helping me cope with the extreme confusion of all the various LK-99 attempts. Like, it's going to be a hugely messy confusion and slog with lots of results that are "blend-A can meet 5 of the 7 requirements, but not the important ones. blend-B can meet 3 of the 7, but one of the important ones. blend-C looks very promising but doesn't quite meet any of the requirements." for... years.
Ignition! By John D. Clark
https://www.amazon.com/Ignition-audiobook/dp/B07CTW1M9D/ref=...
Or, maybe you’ve already read it, in which case: great job. Maybe someone else will see this comment and enjoy it.
I don’t want to spoil if for you, but I loved the last paragraph. I believe a new heroic age for the field is dawning.
Maybe AI to reformat my stylus-scribbles into fully realized edits at the correct location and nicely formatted diagrams correctly blocked into the text. That would be killer.
A little off-topic, but seeing the confusion about (((parenthesis))) used in an unfamiliar context, it's worth mentioning another area of confusion that will eventually surface with SD prompts.
The text-to-image models have biases, of course, and so prompts (positive and negative) need to be used to combat those biases if you don't want the output to go in the biased direction. e.g. if the model tends to draw hats on people, you might need to add the negative prompt ((hat)) meaning "really, really, no hat". Makes sense.
Now, consider when the text-to-image model has a bias regarding drawing people of a particular race, the person using the model may need to write a prompt counter-biasing the result to get the result they want. For example: by saying things like prompt (hispanic) or negative-prompt ((blue eyes)).
Not saying that such prompts are guaranteed not-racist, just that an uninformed observer could naturally suspect racist intent with race-specific tags in the prompt. It's complex, and would need more context to confirm a suspicion.