4 ms·
What about US Gen-Z and younger? They fall into the Oceania all-ages bucket? Heh, just a note on https://news.ycombinator.com/item?id=46541796 https://news.yco
by aebtebeten 9mo ago
What about US Gen-Z and younger? They fall into the Oceania all-ages bucket?
Heh, just a note on https://news.ycombinator.com/item?id=46541796 https://news.ycombinator.com/item?id=46541796 : imagine a therapist believing that the proper route is to help someone cognitively figure it out by asking the right questions! Almost as difficult as imagining a dev who believes the proper route is to write a script that does it... la déformation professionelle?
Wrt: "once you know the right 20-30 people or their friends or colleagues" it makes me wonder if a good corruption coefficient might be "how many zeros does it take[0] before you stop seeing impartial institutions and start seeing corruption". I believe those 20-30 people exist in almost all countries, it's just in some lucky countries you don't notice them at USD 6-7 zeros, but do start noticing them[1] at 9-10 — and in some other countries, you even have the 20-30 people in the quarter who are corruptible for 2-3 zeros.
[0] or maybe an even better index would be calibrated to wealth distribution within the jurisdiction?
[1] ask anyone investing in US offshore wind about US institutional impartiality these days!
Q. Should I read a translation of "Osudy dobrého vojáka Švejka za světové války" sometime?
- gsf_emergency_6 9mo agoI looked at anime related comment trees :) https://hn.algolia.com/?dateRange=pastYear&page=0&prefix=true&query=%22One%20punch%20man%22&sort=byDate&type=comment https://hn.algolia.com/?dateRange=pastYear&page=0&prefix=tru... Wrt the rest: a few more cycles of reflections needed.. (There's public evidence here that 9-10s won't unearth more than the 20 that any Joe knows about. The other 10 spurious ones depend on the situation and mission. I expect the situation in CH is almost the same, via a different design) Compare with HN :)
- aebtebeten 9mo agoI realised this morning I still haven't watched all the Krtečeks yet; should probably do that before delving deeper...
- gsf_emergency_6 9mo agoI surmise that you already have a general rec in mind so I'm just reiterating here for your review: Point the user^W dev towards better "tools of thought" that decouple problem-refinements from solutions. TLA+, proof assistants, AIDEs, etc Outside HN, I've been mostly refining my examples "by hand".. (Hope you guys made it to the CNSOER prequals but if not, I suspect you already know the 10 guys that you need to call :)