bo1024

80 ms·

1. 811.▲
   by bo1024 6y ago

   Sorry to hear you're dealing with this. I'm not in the field, but this is a case where I would abstractly be tempted to use javascript blockchain mining or similarly require some amount of useless computation by the browser during
2. 812.▲
   by bo1024 6y ago

   This falls under fingerprinting and privacy-invasive methods.
3. 813.▲
   by bo1024 6y ago

   I wouldn’t agree. It’s the best communication format for someone who wants to control their own time and attention.
4. 814.▲
   by bo1024 6y ago

   I think he explains pretty clearly up front why his job makes this circumstance appropriate - on the bottom of things. I can’t think of anyone who should be more on top of things than a president.
5. 815.▲
   by bo1024 6y ago

   If you show people two routes, they will choose the faster one. Even if you tell them it would be very helpful to take the slower one.
6. 816.▲
   by bo1024 6y ago

   So it sounds like you want email, or a ticketing system. Not live chat. am I right?
7. 817.▲
   by bo1024 6y ago

   I haven't found any interface I really like, although for fun I rolled my own (along the lines of mutt, which I haven't really tried).
8. 818.▲
   by bo1024 6y ago

   I agree there is a separate but interacting issue that email UIs are mostly very bad, so people like to avoid email.
9. 819.▲
   by bo1024 6y ago

   This raised the question of - why not use email instead, then? Then answer I came up with is that the asker wants an immediate-ish response. But in this case, the "hello" makes sense not as a pleasantry, but as a way of finding ou
10. 820.▲
    by bo1024 6y ago

    This post contains a bunch of information about the question, but it doesn't seem to actually address the question. The question is: does Chromium send the first word I type to my ISP? The answer appears to be: yes.
11. 821.▲
    by bo1024 6y ago

    Ah, this is a nice point, it does work for the number of edge pieces. In this case (3x3x...x3) every piece but one is an edge piece! More generally, with k pieces on a side, there are k^N pieces total, and (k-2)^N internal pieces, so in the
12. 822.▲
    by bo1024 6y ago

    Ha. That sounded close to me but it's not right, I think the reason is that the corners are very small. Suppose a perfect grid in a hypercube. In d dimensions we have 2^d corner pieces. But if there are k pieces per side, then k^d tota
13. 823.▲
    by bo1024 6y ago

    I didn't remember the name, but when I saw the title of this post I was expecting something like write.as. Start typing and hit a button.
14. 824.▲
    by bo1024 6y ago

    Differential privacy can definitely do much fancier stuff. A big challenge is that you like to play around with your data and try lots of different analyses and regressions, and DP tends to clash with that approach. It's more natural t
15. 825.▲
    by bo1024 6y ago

    I think "behind" is not the right way to put it. It only has the basic functionalities, but these are crucial and the implementation looks state-of-the-art to me (at a glance). It doesn't have the fancier machine-learning alg
16. 826.▲
    by bo1024 6y ago

    Great insight, this is a general issue with making decisions based on predictions. It comes up in Hanson's "Futarchy" / decision markets too. For example, if you have a huge bias against recruiters and never take their c
17. 827.▲
    by bo1024 6y ago

    Amazing how end-to-end encryption would solve all the problems they claim to care about here.
18. 828.▲
    by bo1024 6y ago

    Do you have trouble with xournal?
19. 829.▲
    by bo1024 6y ago

    This reply comes from completely internalizing the perspective the above post describes. It only makes sense if you accept that the employer is supposed to completely care for their employees. But if you view it as society’s job, then every
20. 830.▲
    by bo1024 6y ago

    Right, but the chance of this happening is a lot smaller. Actually, I should have realized, what one should do is draw a random Gaussian (or Laplace) with very large variance -- see "differential privacy".
21. 831.▲
    by bo1024 6y ago

    Yeah, you can make this a lot better by just making the random number in the range 1 billion - 1 trillion.
22. 832.▲
    by bo1024 6y ago

    Interesting. But can you distinguish between the effects of contextual advertising (ad depends on the content around it) and targeted advertising (ad depends on the person viewing it)?
23. 833.▲
    by bo1024 6y ago

    I believe Stallman uses, or at least used, "cracker" for criminal hacker.
24. 834.▲
    by bo1024 6y ago

    Don't forget cash.
25. 835.▲
    by bo1024 6y ago

    I agree with a lot of this, but I think there is a consistent pattern of AI announcements playing on humans' intuitions to create the impression that much more has been achieved than can actually be proven -- in fact, not even trying t
26. 836.▲
    by bo1024 6y ago

    Thanks for your reply. A couple responses to advance the conversation. As a side note, it's worth mentioning that apparently, from other responses, it seems we have little idea how much arithmetic GPT-3 has learned, and it may not be m
27. 837.▲
    by bo1024 6y ago

    > We aren't sure, but the speculation is that in the process of training, GPT-3 found that the best strategy to correctly predicting the continuation of arithmetic expressions was to figure out the rules of basic arithmetic and enc
28. 838.▲
    by bo1024 6y ago

    I think this is a very good take.
29. 839.▲
    by bo1024 6y ago

    > Knee implants Source? I would expect these are much more common in obese people than in habitual runners. My understanding is that runners have been found to have healthy knees, e.g. with a bit of googling I find [1][2] (but haven&#x27
30. 840.▲
    by bo1024 6y ago

    I think it’s fair. Nobody thinks a faster algorithm than n log n will ever be possible. It’s like a mini P!=NP.