It was pretty well established scientifically in 1900 that increasing atmospheric CO2 would result in increasing global temperature, but I don't think it was really in the public awareness for many decades. "Global warming" wasn't coined until the '70s.
Right. The equivalent in handwritten code would be formatting your code by hand. That used to be the normal way to do it!
For handwritten code, the evolution of best practices has tended to be:
- just make your code look neat and tidy;
- follow the style and conventions of existing code;
- follow a strict formatting style guide;
- format automatically using a tool.
I don’t see why it should be any different with LLMs. Why format with an LLM each time, when you can use the LLM once to write the formatter?
Maybe there’s a point at which neural nets replace conventional programming languages for low-level tasks. But I’m skeptical that natural language models will replace programming languages for low-level tasks any time soon.
> Why format with an LLM each time, when you can use the LLM once to write the formatter?
the right way to use an llm imo (and same for end-user features as well)... no need to waste tokens and wait time on something that can be done at a fraction of the cost and time (and be deterministic on top of that)
That had me bogling too. But you know what? A local MoE model roughly equivalent to sonnet mid-2025? Totally possible. Just costs electricity to run, put it in your CI/CD pipeline. Have it apply a bit of intelligence to the thing as well. Uh.... if you've got a spare box, why not?
(the fact that said spare box would cost an arm and a leg in 2026 is... a minor detail)
Why not? Because you can get stronger guarantees of correctness and consistency out of a typical code formatter, which will also probably run about a million times faster.
Just last night, I wanted to find some antonyms of a word. So I did what I've done for decades and simply Googled that.
It insisted that I meant synonym, not antonym. Let that sink in for a moment.
Irrevocably substituting the antonym of antonym is the most balls-up, backwards, paradoxical "I'm from Google, and I'm here to help!" thing I can imagine happening to one word.
The quotes did nothing. The search results were all for synonyms, with the word synonym bolded in each excerpt.
---
(Hey, Google: It's fine to present to the user a suggestion, or a correction. I can even work with a system that assumes a correction is good and uses it on the first pass -- I might not like the extra step, but I'll get over it. Sometimes, that's actually useful.
But when your systems present a line that asks "Did you mean 'synonym'?" and then offers no option for the user to -- you know -- actually answer that question and reject the correction, then that's not good.
In fact, some descriptors that come to mind before "not good" in this context are "callous," "insulting," "recalcitrant," and "sadistic.")
There's another more hidden tool avail: right of the search type bar (images/news/books) there is a "search tools" menu where you can open "all results" and switch it to "verbatim".
Often times a good way to see another defunct relict of old, quality google: the empty-result-troll that would once upon a time pull out his fishing rod on click..
A tangent, but this is the second time in two days I've seen the word spelled "often times" instead of "oftentimes". Is this some variant spelling I don't know of? I see it more than "oftentimes" now, which I was hitherto convinced was the only correct spelling.
I believe you may be correct, but they're both readable-enough.
Like "cannot" vs "can not": One form may be more-correct, but both are very readable.
Either way, it's easy enough to blame spell check on our personal pocket supercomputers for these things.
(Every year or two, Google Keyboard on Android makes it its purpose to screw up "its" vs "it's". You type it the right way, you see it on the screen as being correct, and then it changes it to the wrong form. This happens 100% of the time and then the problem disappears in a few weeks.
I'd give Google a break, but they don't deserve one.
I also blame them single-handedly for the variations in spellings of brake-vs-break on the longer timeline: Sometimes, people get it right and nobody notices. Oftentimes, it's all backwards. The oscillation suggests that it is an auto-derp problem more than it is a cognitive one.)
Cannot and can not are slightly different in that both are correct (in the prescriptivist sense, I suppose; arguably whatever gets the point across is correct). But there are cases where can not is more correct.
I use a keyboard (Thumb-Key to be precise) without autocorrect, though it doesn't stop me from making typing mistakes.
I (usually!) want to find documents that include the words that I'm searching for, not an endless stream of links that some particularly-useless bot thinks I might want instead.
(And when that search returns no results, then that is also a useful data point for me.)
For instance, searching the quoted (random phrase) "pants butler" produces first page results like:
"pants,” Butler" and "pants...Butler" and "Pants - Butler's"
Second page loses it entirely, with results like "BUTLER SVC Green Back Country Cargo Pants" and another that seemingly lacks "butler" anywhere on the page.
I have also noticed this. Many other search engines have started doing it too.
If I had to guess, they are probably deferring to autocorrect if a quoted search doesn’t appear often enough to be notable and the distance to existing common tokens is small. This really sucks, because it means that you can’t search for uncommon things that are named similarly to common terms. Once upon a time it wasn’t like this.
A similar problem comes up if you want to clarify a common search with an uncommon term, like (made up example here) “German castle Tokyo”. Once upon a time you could quote the uncommon term or prefix it with a plus to force a narrowing of the results. This could find discussions or specific posts with unusual combinations of words, which was great when you knew were looking for something very specific and obscure. Now this hardly ever works, and instead they just ignore your extra term.
Sometimes the search engine “AI assistants” can find these things if you prompt correctly, which is maybe the most useful application of AI that I’ve found. But even then they often don’t seem to search that deeply, and often they will just assume that your query is invalid and gaslight you.
Even if the Apollo program was similarly politically motivated, it at least was seriously cutting edge science and engineering. I mean, there were many people born before the Wright brothers’ first flight watching the moon landing on TV. Basically repeating Apollo 8’s much less iconic flyby decades later is obviously going to be less impressive.
Ok, but society can't bear the cost of $10M garbage men, so either people will do it themselves or go without.
The same argument applies to any job: in most scenarios, it pays what it's worth to society at the market clearing price. The government can interfere via licensing, minimum wages, quotas, etc; but broadly the job pays what it's worth.
Pay and worth are different, just like price and value are. Garbage collection is worth a lot, but its pay is determined by market dynamics. As the number of unemployed increases, it will pay less.
You left out “at the market clearing price”, then described a scenario where the supply of labor increases and the price drops, proving my exact point.
“Worth” and “at the market clearing price” are different concepts, which was my point. The job doesn't pay what it's worth, it pays the market price, which is determined by market dynamics, not how much people value it.
Society cannot function without garbage people. It must be done or society will collapse quickly, so society better find a pay rate that gets the job done.
This thread feels like it went off the rails given that every locale I've ever lived in (many across the US) had fine, working garbage collection, and plenty of competent garbage men who worked for what I'm guessing was decent pay, certainly less than $10 million a year.
I think in reality, it shouldn't be hard to find people willing to take out garbage by simply paying a little better than other manual-labor jobs. There's always going to be people who can't work other jobs for some reason, so if they're choosing between manual labor jobs, the one that pays more is going to be more attractive. They don't need to pay enough to hire a doctor, because not that many people can do high-value work like that competently.
And there will always be people who think that driving around in a heated/air-conditioned truck with a big claw that picks up and empties the cans is a fun job and will happily do it at any reasonable pay level.
Not everyone wants to work in an office; some people really like working outside and being indoors is like prison.
Those jobs require some physical handling of the garbage. That big claw doesn't always work as expected, and when things go wrong, they have to get out and clean up the mess. There's also items that don't fit in the can and have to be manually handled.
But you're right, the job isn't as bad as it used to be.
Well, my point wasn't really that the job has gotten easier, it's that some people simply prefer doing that kind of work.
I remember being in a meeting once when a manager looked out at the guys on a scaffold cleaning the windows and remarked "at least we don't have to be out there like them." My first thought was that they were probably thinking "at least we don't have to be inside like them."
While I can understand the sentiment, I think most of this is fantasy. A lot of people vastly underestimate the toll manual labor takes on one's body. It may be nice to ponder "Wow, it must be so great to work out in the sunshine all day!", compare anyone doing physical labor (and I count extended driving in that) age 50+ with any office worker 50+. Sure, anyone 50+ has standard aches and pains, but I've seen many physical laborers dealing with constant pain and a lot of degenerative diseases at that age.
The thread started out off the rails. Contrary to the claims of youre-wrong3, garbage collection is not a particularly high paying job and has no real trouble getting new hires.
I'm not sure you'll attract the right kind of people with that much money. Probably a lot of people looking for a quick buck will apply, do a good job for long enough to not get outright fired, collect a few cash checks, and then knock it off.
I think there would actually be enough people willing to do blue collar jobs if job security is alright. Low-status jobs are defined by low job security, not necessarily by harsh working conditions.
reply