Semaphor

42 ms·

901.
by Semaphor 3y ago
> Note that the VSL [value of a statistical life] is very different from the value of an actual life. The part about the calculation the EPA uses is very relevant: > Suppose each person in a sample of 100,000 people were asked how muc
902.
by Semaphor 3y ago
Last time in July 2021, 89 comments: https://news.ycombinator.com/item?id=27758962
903.
by Semaphor 3y ago
Seems like a bad adblocker, uBo works as expected
904.
by Semaphor 3y ago
I don’t think Binance allows you to do much with just a password. Not that the "theory" really needed more debunking.
905.
by Semaphor 3y ago
Amazon (sometimes I really want some music not on a proper site like BC) does it with music they have the digital rights for (and some they don't). They also do weird things like selling CDs that include the MP3s for cheaper than just
906.
by Semaphor 3y ago
Yeah, during COVID, I rented a car both times for Christmas, hated it, glad I'm now back to trains.
907.
by Semaphor 3y ago
Not that I know, only family sharing. My wife and I have the same issue
908.
by Semaphor 3y ago
So that’s what that does. I have to press the FN key as modifier, and it does indeed open the menu. Never felt the need to press it before ;)
909.
by Semaphor 3y ago
It seems around 0.6€ is where they judge it worthwhile to create an invoice. I have even less data (because I only backup my server there, for my PC I use backblaze personal with payments every 2 years), and they charge me 0.6-0.85€ every 2
910.
by Semaphor 3y ago
Only if they changed that (which they might have as part of their cost-optimization). They said they mixed bing and google results back then.
911.
by Semaphor 3y ago
I’m clear why they didn’t include it, I wanted them to tell me why, though. And I thought that both of them apologized in almost the same way, was funny.
912.
by Semaphor 3y ago
Kagi mixes google, bing, some non-profit small-web SE, and their own index.
913.
by Semaphor 3y ago
With a better prompt, you can get it to list some, but it’s very annoying to do so. Mistral showed that their medium model is far better (yet not good), and the same prompt as in the article gives only one instead of 3 paragraphs of ramblin
914.
by Semaphor 3y ago
What region? I get similarly bad results with international (and a quick check with region US also didn’t improve things) and uBo at only #5, and ytdl at #12. And I already have github on "raise" and a bunch of domains blocked (no
915.
by Semaphor 3y ago
Besides what sibling said, and lack of bloatware, I find LineageOS to simply be better than original ROMs. Just using vol up/down to move the cursor is already a godsend for me. I’ve also been using CyanogenMod/LineageOS since my
916.
by Semaphor 3y ago
It’s not about hard/easy, it’s the supported devices. I spend very little on phones (current one is a OnePlus Nord, which I bought for 250€) as I don’t like to use them.
917.
by Semaphor 3y ago
I unlock the bootloader and install a custom rom (pretty much exclusively LineageOS nowadays, I’d try GrapheneOS, but I’d have to spend far more on the phone than I’m willing to get that). That is everything cleaned ;) I only buy phones wit
918.
by Semaphor 3y ago
That explains it, I was wondering why things suddenly worked. Not just relevant for devs, but also people who want a better and/or more private android experience than what vendors offer.
919.
by Semaphor 3y ago
They mention it a lot, and again at the end in the cautionry section: > And for the love of all that is good, if you actually need a tree, use a tree. It’s in almost every section.
920.
by Semaphor 3y ago
It wouldn’t, we’ve been planning to make the site mobile-friendly every year for 3 years now, maybe 2024 it’ll happen :D Let’s just say the way the site works on mobile does not give us as many mobile visitors as you’d expect ;)
921.
by Semaphor 3y ago
Been using it since quantum, it happens after some updates, recently there was some modal about some feature I didn’t care about, and I had to click twice to close it. That was only a few weeks ago.
922.
by Semaphor 3y ago
Have you tried a user-agent switcher? I think you can even set them up to always have a specific agent depending on the domain. Chances are, they don’t need chromium, they are just written by (even more than you’d think) incompetent people.
923.
by Semaphor 3y ago
That’d be interesting to know, but I don’t use anything apple.
924.
by Semaphor 3y ago
It will have to stay a mystery, looking at distribution, both algorithms looked the same.
925.
by Semaphor 3y ago
I exclusively use FF, with Chrome on Android and Desktop only being there for cross browser testing, I never have rendering issues, only issues brought by extensions (mainly uBlock Origin), or FF’s tracking protection, both of which can be
926.
by Semaphor 3y ago
I posted some stats from our website a few weeks ago [0], GA was heavily undercounting FF compared to our server-side stats based on UAs. One other thing to remember, is to check the falling of desktop usage, because a large part of the mod
927.
by Semaphor 3y ago
Okay, now you made me want actually look into that ;) I should be able to recreate the old code (it was years ago, but I still have the script) and make a comment if I end up actually finding statistical anomalies.
928.
by Semaphor 3y ago
The old code was var r = new Random() ;) I'm assuming I either hit an PRNG bug (after all, I encountered a compiler bug in university already), or more likely, that I simply imagined the issue or bad bad luck with the data. Never did
929.
by Semaphor 3y ago
I once wrote a small script (in C#) to pick a few tens out of a few 1000s. I got weird repeats. I switched to a cryptographically secure RNG, and the repeats were gone. It was probably pure chance, but I stopped using the normal random func
930.
by Semaphor 3y ago
Considering I expected it to be about LLMs from the title (I thought it would be an announcement of a new 4B model), I’m going with "probably" ;)

More ›