bcoates

10 ms·

91.
by bcoates 1y ago
He was always long-winded and discursive (in the needs-and-editor way, not the interesting way) but I'd actually enjoyed his earlier works (this one is pretty popular: https://www.ribbonfarm.com/2009/10/07&#x2
92.
by bcoates 1y ago
Imho, this is wrong. Even independent of access to vast amounts of compute, symbolic methods seem to consistently underperform statistical/numerical ones across a wide variety of domains. I can't help but think that there's m
93.
by bcoates 1y ago
California Assembly, and they do have that power, and they just used that power and the answer was "lol no we don't care if you want to waste your own money that's fine"
94.
by bcoates 1y ago
The system works? The original bill was a going-nowhere, feel-good sop for chumps, the chumps were placated, the bill served its purpose, everyone moved on. Imagine a world where random crusading busybodies were actually given the power to
95.
by bcoates 1y ago
I mean... most businesses, particularly small businesses and startups, aren't exactly doing brain surgery on a rocketship. It’s pretty likely that they have extremely dull problems like "running an inbound call center is a lot of
96.
by bcoates 1y ago
Yep. I'm looking forward to LLMs/deepnets being considered a standard GOFAI technique with uses and limitations and not "we asked the God we're building to draw us a picture of a gun and then it did and we got scared&quo
97.
by bcoates 1y ago
I think the productivity gains of dynamic typed languages were real, and based on two things: dynamic typing (can) provide certain safety properties trivially, and dynamic typing neatly kills off the utterly inadequate type systems found in
98.
by bcoates 2y ago
Oh, I vastly prefer people who don’t understand and know it. Reminds me of Frank Zappa comparing "cigar chomping old guys" to the "hip young types" that replaced them https://youtube.com/watch?v=KZazEM8cg
99.
by bcoates 2y ago
This is a nit, but grocery bagging (one of the article's examples) is a no-win situation. I worked it as a summer job a while back--It’s the most bikeshed job in existence because (nearly) everyone understands it, but they all have the
100.
by bcoates 2y ago
Clever comment by the illustrator in the headline image: https://genius.com/Gil-scott-heron-whitey-on-the-moon-annota...
101.
by bcoates 2y ago
(Only) the process is elevated, but the process has a window on a shared session, and the OS does not successfully protect processes that share a session (and user, and registry, and disk, etc., etc.) from controlling each other. From an AP
102.
by bcoates 2y ago
Windows has a whole different (looser, older) security model. There are no security barriers between windows running on the same desktop. (In particular, "UAC is [still] not a security barrier"--when you hit ok/type in a pass
103.
by bcoates 2y ago
Do the copyright claims have any legs at all? ianal, but I thought it was pretty settled that statistical compilation of copyrighted works (indexes, concordances, summaries, full-text search databases) were considered "facts" and
104.
by bcoates 2y ago
Looks like alternativeto.net wins the most divisive prize, for being on the top boosted list but also heavily banned/demoted. Runner up to the NYT
105.
by bcoates 2y ago
Tiktok has a lot of deep-ish search keyword spam that doesn't actually lead to anything related to the query, just some generic videos and a pitch to install the app. It's basically useless as a website.
106.
by bcoates 2y ago
I'm not trying to pick on you or anything, but at the top of the thread you said "I mean, I can ask for obscure things with subtle nuance where I misspell words and mess up my question and it figures it out" and now you'
107.
by bcoates 2y ago
Apparently GPT thinks the earth is flat.
108.
by bcoates 2y ago
All the good stuff was in Area51
109.
by bcoates 2y ago
And yet, the silver report final modal outcome was exactly the actual result.
110.
by bcoates 2y ago
Innumerates keep self identifying, even 8+ years later. Please learn anything the Fremdschämen is getting painful.
111.
by bcoates 2y ago
I don't have a published study to back this up, but in my 10 < n < 100 experience, the alcohol suppression effect is limited to occasional-to-moderate drinkers. Giving a GLP-1 to a heavy drinker/alcoholic results in a skinny
112.
by bcoates 2y ago
That seems to be a problem specific to PySpark, it’s amazing how poor search results are for it despite it being reasonably popular.
113.
by bcoates 2y ago
It's just SF, there's fiber-to-the-pole or better in most of the LA area, even if the only last-foot service is DSL or cable
114.
by bcoates 2y ago
Sure, but I just use OneNote, the company wiki, or like physical sticky notes for that. I'm not seeing the value of having the LLM give feedback aside from entertainment
115.
by bcoates 2y ago
I dunno, his cake baking dog looks pretty honest to me.
116.
by bcoates 2y ago
On larger codebases, I use tools heavily instead of just winging it. If the LLM can't either orchestrate those tools or apply its own whole-program analysis it quite impossible for it to do anything useful.
117.
by bcoates 2y ago
That lonely/downtime section at the end is a giant red flag for me. It looks like the sort of nonproductive yak-shaving you do when you're stuck or avoiding an unpleasant task--coasting, fooling around incrementally with your LLM
118.
by bcoates 2y ago
Very common in light commercial applications but almost unheard of in residential because nobody installs nema 20a sockets in houses even if the circuit can handle it
119.
by bcoates 2y ago
Diet interventions aren't terribly hard to test, they just consistently tend to result in "as bad or worse than no intervention" (short of fixing the sort of malnutrition you rarely see in the contemporary developed world) Th
120.
by bcoates 2y ago
Given how absolutely trivial it is to do a randomized intervention study here (you can't do double-blind, but dose-response is easy enough) there's not a lot of reason to care about n=1 results except plausible reports of rare s

More ›