You'd think so, but I've seen plenty of very good answers get downvotes as well, and if you look in the voting records you'll see certain SO accounts that simply carpet-bomb every question and every answer with a downvote, reputation costs be damned.

Why they do that, I don't know. Some weird form of trolling?

It might be a personal problem, but I suspect it might also depend on the category of questions the questioner asks. For example, a person asking a C++ question on StackOverflow might have a very different experience from a person asking a CSS question on StackOverflow, simply because the development communities associated with the two languages have different cultural assumptions and thus attract different types of answer-ers.

You're not wrong, but StackOverflow's methodology for handling this problem proved (in hindsight) to be inadequate, because it maintained the experience quality for established users at the expense of new users, and a site like StackOverflow needs both kinds of users to thrive.

A better mechanism might have been to allow repetitive newbie questions, allow people to answer them as well as they care to, and then have an asynchronous "garbage collection" background process (either human-based or automated) that digests the redundant newbie questions into improvements on the canonical ones, and/or collates them into a second tier of non-canonical questions that are deprioritized in the search results. That way the newbies get the help they are looking for ASAP (which is what will bring them back) rather than the pain of rejection, while the experienced users get a well-organized, non-redundant site experience (ditto).

Life is WAY better after the industrial revolution than it was before it.

People have this fantasy image of what life used to be like, thinking of picturesque farms, craftsmen tinkering in workshops, clean air, etc. The middle ages were filth, you worked backbreaking labour long hours of the day, commonly in highly risky environments, even the simplest necessities cost a large portion of your salary, you lived in a hovel, and you died of preventable diseases at an average age of ~35 (a number admittedly dragged down by the fact that 1/4th of children didn't even survive a single year).

If it takes people of similar social status as you weeks of labour to produce the fibre for a set of clothes, spin it into yarn, dye it, weave it, and sew it, then guess what? It requires that plus taxes and profit weeks of your labour to be able to afford that set of clothes (and you better believe the upper classes were squeezing every ounce of profit from the lower class they could back then). Decreasing the amount of human labour needed to produce things is an immensely good thing. Furthermore, where did that freed up labour go? Into science, into medicine, into the arts, etc etc. Further improving people's quality of life.

And if your response is "But greater production is more polluting!" - I'm sorry, do you have any understanding of how *miserably* polluted cities in the middle ages were? Where coal smoke poured out with no pollution controls, sewage ran straight into rivers that people collected water from and bathed in, where people extensively used things like arsenic and mercury and lead and asbestos, etc etc? The freed-up labour brought about by the industrial revolution allowed us to *learn* and to *fix problems*.

Meh, you do you. I'm rooting for the Torment Nexus.

Which of those things hit 800 million users in 17 months?
Which of those things hit such high annual recurring revenue rates so fast?
Which of those saw the cost of using the tech decline by 99% over two years?

Heck, most of them aren't even new technologies, just new products, often just the latest version of older, already-commonly-used products.

And re. that last one: it must be stressed that for the "cost of using the tech" to decline 99% over two years per million tokens, you're also talking about a similar order of reduction of power consumption per million tokens, since your two main costs are hardware and electricity.

You're looking at three months of very noisy data and drawing some pretty dramatic conclusions from said minimal data.

Winter demand is heavily dependent on weather. You're mainly seeing the impacts of weather on demand.

https://ancillary-proxy.atarimworker.io?url=https%3A%2F%2Fen.wikipedia.org%2Fwiki%2F2024%25E2%2580%259325_North_American_winter

"The 2024–25 North American winter was considerably colder then the previous winter season, and much more wintry across the North American continent, signified by several rounds of bitterly cold temperatures occurring."

Probably a woke lunatic who believes in a spherical earth and heliocentrism. Doubtless the Administration has a Pentecostal pastor who believes in Creationism and the healing power of CO2 to run the agency.

... at which point, the company's costs will drop to zero; making their cost-cutting initiative a complete success!

I have definitely produced useful SQL code, and indeed some pretty darned complex queries for transformations and data hygiene, but as I said, it's not a process of "dump spec into LLM model, run SQL on RDBMS", but rather a kind of meta-programming conversation. I imagine specialized LLMs might do a bit better, but I'm generally pretty skeptical of the current generations of AI building sophisticated software. I suspect where LLM's might do well is with interop code, the kind of boiler plate code that takes up a good deal of a programmer's time. That's likely why I have had some success with SQL.

My experience, mainly with generating SQL queries, is that AI inevitably gets it wrong multiple times, so what I have had to do is more of a kind of meta-programming; giving the model cues and corrections. I have created some pretty sophisticated SQL queries, but there's no way in hell I can just pop the first go-around into my code and have it run. Either it's outright faulty code that will fail, or it's just not producing the correct results.

Now SQL is a fairly limited and ring fenced language (excluding stored procedures of course). I've never tried it with a general use language, but I imagine those problems will get more pronounced. That's not to say it might not be useful for translating natural language specs into code, but if my experience with SQL is any indicator, it's going to require a lot of massaging. There's probably still productivity boosts to be found here, which will likely have in effect on the number of programmers out there, but to me, it feels more like a layer of abstraction that will require a different kind of programming, rather than replace programming.

As an example that isn't coding, I have been building models for what I expect is a government procurement next year. This involves taking previous Requests For Qualifications documents, updating them with current knowledge of government expectations, procurement rules, and so forth. Again, building these model RFQs is an iterative process, not simply one of "Take these RFQs from previous procurements, update them with this new information I've uploaded, and give me model RFQs based on these premises I will provide." My test run took about three or four hours of a kind of conversation, where I correct and shape, understanding the cues the LLM needs to produce the desired result, and the better I get at understanding not just the kind of information and cues the LLM requires, but the most effective means of "encoding" that information, the more efficient the LLM is at producing the desired results.

That sure sounds like programming to me, albeit at a much higher level of abstraction. LLM, at least where it stands, is just another platform, a very powerful one, but as with all programming languages, the larger the command set and the more complex the lexical structures, the more room for bugs, and the more subtle some of those bugs can be.

Instead of letting an LLM run a business directly, and hoping it will continue doing the right thing indefinitely, how about:

1. Task an LLM to write a state-machine/rule-set for running a business (e.g. lots of "if X happens, do Y" rules)
2. Have a human review the state-machine
3. Have a computer run the state machine (which is very unlikely to go haywire since it isn't an AI, rather just a set of rules)
4. If/when the state machine proves insufficient in some new scenario, goto (1)

At some point this is going to reach "Open the pod bay door Hal" territory.

So you do deny thermodynamics

Every "solution" the skeptics lay claim to requires obscene amounts of energy. Whether it's desalination or carbon sequestration, they would require incredible amounts of energy to be at all scalable. Which leads us to the great paradox of the quasi-skeptics and their "build more shit to absorb the shit we're emitting" pseudoscience. If you can produce so much energy that you could do large scale desalination, effectively sequester those salts, other minerals and metals so you don't kill off large proportions of marine and land ecospheres, then you've already solve all the important problems that cause the climate change effects, and you don't need to do anything other than mitigate what would be a few centuries of ill-effects of previous emissions.

Every time I see these kinds of claims, it just baffles me. I had one guy tell me that Alberta could keep pumping out bitumen and natural gas, because nuclear, wind and solar would provide enough energy to capture and sequester all the GHGs. He seemed absolutely baffled when I informed him that thermodynamics put such high constraints on the efficiency of any such mechanism that energy requirements would exceed (likely by an order or two of magnitude) any energy extracted directly from the fossil fuels themselves, and further, you wouldn't even need those fossil fuels any more.

Basically the skeptics, quasi-skeptics and the climate nihilists (those that accept we're doing substantial harm, but just don't want to do anything about it), are either completely ignorant of physics, or simple don't give a damn and have effectively become hedonists.