repsilat

29 ms·

271.
by repsilat 7y ago
> which lets you write macros using Python Someone upthread said "nocode", and I think that's more on the money. Most people will never be "programmers" as we understand the term, but they could still benefit f
272.
by repsilat 7y ago
I think so. Starlink is capital-intensive and untested as a business proposition. This keeps the risk walled off from SpaceX and lets them raise without diluting ownership of SpaceX. (It also means guaranteed profit for SpaceX from the pock
273.
by repsilat 7y ago
Hough transforms are "old-school" computer vision. They rely on a know parameter-space, "This is a circle, that is a straight line" and that isn't super robust to noise or to unfamiliar curves. I don't know for
274.
by repsilat 7y ago
I think this sort of statement is thrown around too lightly. It's a bad sign, but there's a limited amount of signal in it. It may not correlate strongly with employee experience, and lots of other positive factors might outweigh
275.
by repsilat 7y ago
> ...if you don't like me and my solutions, what else is there? I'd like to broaden this position a bit before laying out the alternative. I think the philosophy here (which I largely share) is that "if what you're
276.
by repsilat 7y ago
The former is incorrect, the latter merely slower than necessary.
277.
by repsilat 7y ago
Right -- today you need to remember to charge your phone, you don't take it everywhere (and don't have signal everywhere, especially internationally), and you need to take it out of your pocket to use it, and type into it with you
278.
by repsilat 7y ago
There's a bit of a push to get everything that isn't rendering out of the render thread, and if you're going to do a bunch of logic in a web worker I guess it isn't a big jump to do it in Go/Rust/whatever. I ag
279.
by repsilat 7y ago
That article talks about map and filter and fold, but I think there are also (trickier) ways to make joins and sorts work as well. I think it could be a great technique for a lot of things -- compilers and database views, in addition to fro
280.
by repsilat 7y ago
> Most functional languages have immutable persistent data structures that don't require a linear copy and are fairly efficient. This is true, but unfortunately those libraries/languages don't make use of that tree struc
281.
by repsilat 7y ago
> Anyone have first-hand experience with CreditKarma tax filing? I've used it for the last 3 years in California. Quite simple, free, supports backdoor Roths. Tax returns were quickly accepted and paid out into my bank account. Du
282.
by repsilat 7y ago
One idea I though of in the past: if you have a very high income, your income tax rate can approach 50% (state+federal). If you donate appreciated assets to a DAF, and break the effects down by basis+gains, the numbers look like - 50ish% ba
283.
by repsilat 7y ago
This seems more flexible to me somehow. Using the approach in TFA, can the laser handle tokens that are prefixes of other tokens? Or even tokens that share prefixes? Lexing "truefalse" as two adjacent tokens "true" and &
284.
by repsilat 7y ago
Or a spreadsheet.
285.
by repsilat 7y ago
> this Spectator article somehow managed to avoid even alluding to carbon. It was alluded to in this section (ctrl-f "emissions"): >> A wind farm requires far more concrete and steel than an equivalent system based on
286.
by repsilat 7y ago
Also a googler. > because you think that a good open-source license will not protect you/your work from being stolen (!). My reason for wanting to keep copyright is that I want to be allowed to work on my code under the same name
287.
by repsilat 7y ago
Sorry, I thought your original post was meant to address my original point. I said that the Born Rule (the association between squared amplitudes and subjective probabilities) isn't intuitive in MWI, and I thought you said "It mak
288.
by repsilat 7y ago
This sounds like the most reasonable conclusion until you actually start experimenting. An example taken from another thread: Fire a laser of polarised light at a polarised filter angled at an offset of 30°. Measure the light on the other s
289.
by repsilat 7y ago
I don't think it helps, I think it just punts on the question. Let's get concrete: I fire a polarised photon at a polarised filter oriented 30° off. In Copenhagen there's a 75% chance it goes through, and a 25% chance it hits
290.
by repsilat 7y ago
The "measurement problem" is really a Copenhagen-specific thing -- "measurement" is the fundamental and identifying mechanism of the interpretation, and it's problematic. And In the Copenhagen interpretation the Bor
291.
by repsilat 7y ago
I'm also a Many-Worldser, but I don't quite know how to answer the question "How does the Born Rule work?" That is, if "worlds" branch off with some amplitudes, and there's a "me" on this branch
292.
by repsilat 7y ago
I guess you could argue that Tech and Eng getting their own letter means they don't fit under the "S" in STEM, so STEM might not include the non-enumerated applied sciences. (Nor fields of "applied technology", like
293.
by repsilat 7y ago
> Humans create audience segments Eventually there will be machine learning models involved (maybe there are already? I don't know ads.) Those models might move closer to "perfect targeting" where advertisers could compe
294.
by repsilat 7y ago
It probably varies by company, and by team. In my old hobby code there was a lot of std::map, but in my more recent professional experience I see a lot more absl::flat_hash_map and friends.
295.
by repsilat 7y ago
> Re/Targeted ads require the decision to be made in milliseconds, including network transit time If they're served on page-load when a user first visits a site, but I guess a significant minority are also requested+served in
296.
by repsilat 7y ago
There's also a countervailing force, depending on the market. First an example that supports what everyone is talking about: * User X loves cola Pepsi and Coke both bid more to serve their cola subscription services or whatever. Ad pri
297.
by repsilat 7y ago
A tangentially related fact: on the personal website of the Australian science fiction author Greg Egan, it says > There are no photos of me on the web. https://www.gregegan.net/ Also: http://www.gregegan.ne
298.
by repsilat 7y ago
You're right, this algorithm optimizes for "1 point for picking the best option, 0 for anything else", not knowing the distribution. If you do know the distribution, and have some objective function, and samples are "f
299.
by repsilat 7y ago
It isn't optimal for known distributions. If you're sampling from a normal distribution with known parameters, and the first of 100 samples is 5 standard deviations above the mean, you take it.
300.
by repsilat 7y ago
Funny, I'm sure a lot of people think the opposite, and that only proves your point more.

More ›