scarmig

26 ms·

421.
by scarmig 3y ago
People get endorphin hits from both. The fact that you can consider a handbag a speculative investment that you can resell is gambling every bit as much as actual gambling: the expected value is still negative. With resale, taxes alone imme
422.
by scarmig 3y ago
You could also end up winning $100 from gambling $100. Or, on the other end, you could buy a $50 handbag instead of a $500 one. Both are tossing money into a fire.
423.
by scarmig 3y ago
1000x. Moreover, the best piece of financial advice I ever got was that combined with automatically depositing as much of my paycheck as I could afford into an investment account with auto purchase set up. Simplicity trumps optimality. And
424.
by scarmig 3y ago
It's just entertainment. Is blowing $100 on sports betting really any worse than $500 on a designer handbag? Which isn't to say it's not an issue--both are. But despair driving people to spend increasingly on short-term endor
425.
by scarmig 3y ago
Is it? The actual discrimination can be quantified in a typical way. Attractiveness is only a bit trickier, but there's a high degree of agreement on attractive vs unattractive. Give a team of raters a group of people, present them pai
426.
by scarmig 3y ago
"Read The Fucking Manual." It was a common phrase earlier in the tech ecosystem when someone would come in and ask simple questions that could be answered by reading the man pages, although it has fallen out of favor because it&#x
427.
by scarmig 3y ago
There is some efficiency loss: capital was deployed less efficiently than it would have otherwise been, and that costs money, which means it costs lives. Depending on the particularities of the corruption/fraud, that could be in the mi
428.
by scarmig 3y ago
> if schools are understaffed and teachers are overworked More the norm than the exception. This is still good, though, because the public being unwilling to appropriately fund education is almost a law of nature.
429.
by scarmig 3y ago
Exactly. Where are the replications? Has the culture of research in general (not just in psychology) become so perverse that researchers have no incentive to conduct real research? Unfortunately, I don't see gravestones as offering a g
430.
by scarmig 3y ago
My comment doesn't say that psychological sciences are less real sciences than the real sciences. It's wishing we also had studies to see if even studies in the real sciences are replicable (and, when they aren't, do they sel
431.
by scarmig 3y ago
My main complaint about this: anyone paying attention knows that psychological "sciences" are deeply troubled fields at best. What I want to see is something like this study performed against what we think of as real sciences. Is
432.
by scarmig 3y ago
I'm not defending panpsychism; I'm pointing out that a hand-wavey argument that "consciousness is just something that arises from nervous systems that are ours or very similar systems" doesn't explain or justify it
433.
by scarmig 3y ago
How would you evaluate whether other entities with sophisticated information processing behaviors are conscious or not? Just outright rejecting anything that does information processing on a substrate very dissimilar to ours?
434.
by scarmig 3y ago
> Lately I've been wondering... is this a problem, or a strength? It probably depends. But an idea I've been playing with: because transformers have such a strong ability for recall during inference, they might be introducing a
435.
by scarmig 3y ago
XNOR does not distribute over AND or any other binary operators (try 0 XNOR (0 AND 1)), nor does it have a multiplicative identity, so it's not really multiplication in a way that's useful.
436.
by scarmig 3y ago
It's more like, addition is an XOR. (In fact, AND as mult and XOR as add are GF(2)). Throw in NOT (or, really, just a constant 1) and you can compute any circuit. Biologically, inhibitory neurons are every bit as important as excitator
437.
by scarmig 3y ago
1) That's not how index funds work: they attempt to track a target market index. At most, it happens indirectly: get your company into the major indices by a short-sighted depletion of capital, and you can get some level of lift from i
438.
by scarmig 3y ago
The issue with the OkC data is that self-reports aren't particularly useful. The most meaningful metric is actual conversions (in the OLD context, you could call that either a match, the existence of a reply-response pair, or something
439.
by scarmig 3y ago
The most interesting part of that to me is on selectivity. Male selectivity is invariant to changes in group size, while doubling the group size (speed dating context, something like doubling from 10 to 20 IIRC) makes women increase their s
440.
by scarmig 3y ago
The Bitter Lesson. https://www.cs.utexas.edu/~eunsol/courses/data/bitter_lesson...
441.
by scarmig 3y ago
Key point: Nintendo is also not inherently entitled to copyright protections. We grant it the ability to exercise violence on people who copy Super Mario Maker. It's a perfectly reasonable compromise to require that people who want the
442.
by scarmig 3y ago
You could have some side channel with an identifier for the RPC. If the server receives duplicates of one RPC, it ignores the extras. It would require some client support, though. Alternatively, if the RPCs have the same arguments, the fram
443.
by scarmig 3y ago
Of course, one person's ethical stance is another person's inane diversion. Imagine someone started disrupting a PR event by calling Biden a pedophile or whatever the outrage du jour is. Do they deserve protections?
444.
by scarmig 3y ago
Countermeasure: "AI, add ads please. Make them discreet and not readily identifiable as an ad."
445.
by scarmig 3y ago
Person A: "This thing is a bad thing." Person B: "You're just pointing out it's a bad thing to make the person doing the bad thing look bad!"
446.
by scarmig 3y ago
The sense at Google is that it's dangerous to rock the boat too much when it comes to certain pieties. It would have been a good idea to have someone doing that, and Google certainly has the money to, and plenty of people thought it wa
447.
by scarmig 3y ago
If things can interact, how can we be sure that there is no information being passed around in it? Information processors might still arise, just computing over timescales unimaginable to us. By comparison, you might have information proces
448.
by scarmig 3y ago
If nothing else, this pushes OpenAI to release its next generation in the next couple of months. It can't afford to rest on its laurels.
449.
by scarmig 3y ago
SF is probably the second worst managed organization I interact with on a daily basis. PGE is the worst.
450.
by scarmig 3y ago
A related paper I just found and am digesting: https://arxiv.org/abs/2012.04728 Softmax gives rise to translation symmetry, batch normalization to scale symmetry, homogeneous activations to rescale symmetry. Each of th

More ›