Thank god Wikipedia isn’t run like Stack Overflow. As an end user, they have pretty much the same value proposition: user generated answers to my questions. Wikipedia is still doing well, meanwhile it seems SO is constantly being driven off a cliff by bimbos in management.
Not everything needs to be a damn unicorn. SO is an information repository. They need to accept that stop trying to “enhance” it with more crap because they don’t realize their median user is a junior dev who really just needs to serialize a Java object and isn’t going to pay or put up with any LLM-generated nonsense.
SO doesn’t need large language models. What they really need is a better model of what answers are good, what answers are outdated, and what answers should be expanded to include more info (and sometimes, what answers should be slimmed down a bit). Turn the top answer to popular questions into a wiki so that everyone can update it. And then add backlinks for questions which were closed for being “duplicates”. It solves so many problems SO has.
Another thing. This “comments aren’t for extended discussion” nonsense needs to go too. Any question could easily include a Reddit-style discussion tab to facilitate discussion. I’m sure much of it would be at least as valuable as the answers themselves.
More people needs to understand this. It's fine being a small(ish) business that turns a profit and provides a service that's beneficial to society. You don't need to be a billion dollar company to be important or do great work.
DHH talked about this 15 years ago. https://www.youtube.com/watch?v=0CDXJ6bMkMY
Europe is considered economical failure because there are not enough unicorns, there are lists on Twitter with list of unicorns per country that are supposed to show the decline of Europe. No matter if Europe has some of the best living condition for large group of people.
IMHO this thing is ideological, I even feel uneasy mentioning this because it is something we are not supposed to talk since it can start a flamewar and flamewars are how you get your account restricted.
And importantly: because you are a billion dollar company doesn't mean you bring good to the world. At all.
The obvious solution is a compensation cap, not just because CEO comp has exploded while lower rung compensation has virtually stagnated, but also because it might put an end to the constant drive of companies to just gobble up competitors.
I can’t find it now. But Spolsky himself wrote about how Fogs Creek Software could be what we now call a “lifestyle company” that could grow slowly. But he felt he needed outside investment so Stack Exchange could grow fast since it was only useful if it had network effects.
If there were legal and regulatory pressure that crushed and broke down these behemoths (e.g. a FTC and judges that believed that predatory pricing was real and prosecuted accordingly) the VC model would break and this stuff would stop happening.
> More people needs to understand this.
I sounds more like neither one of you has had that choice. Most people, given the chance, would rather add a zero to their savings rather than “doing the right thing for society”
VC’s aren’t interested in your “smallish business that turns a profit” and neither are the LPs that invest in the fund.
A hundred and fifty years ago people risked incredible suffering in order to strike gold. All when slow and steady profit could be made by owning a farm.
Many tech companies are only plausible with VC funding. In order to get VC funding, you generally have to have a path towards being a unicorn.
Its not possible to run a business that loses $84 million. You will run out of money.
It's fine, but it's no unicorn. And having such a business might be the closest and only chance the people involved have at real wealth and impact - they are 90% there, and only need a little scale, or so they think.
If you were in their shoes, you would do the same.
I have a belief that they're caught in a very bureaucratic "we need to use your budget otherwise it would be put into question", but it also means when I give 1 euro to them it goes less and less to their core mission I want to sustain.
Just stop giving them money, give it to some other project which is also doing something valuable but which needs it more. If it ever turns out the Wikipedia needs your money urgently to keep the servers running, start giving to them again.
Maybe its time for a leaner alternative?
I can't say the same for other enshittified companies.
(Un?)Ironically, one of my main uses of ChatGPT is to replace StackOverflow (it's great at turning vague guesses about what I want into fast ideas, and when it's wrong it's still less wrong than the combination of SO content and the search engines connected to it); and also for turning undocumented and badly documented examples of JSON into a collection of (swift) `Codable` structs.
I'm pretty sure it's the main usage of almost all the programmers who use ChatGPT.
But I don't entirely agree on LLMs being useless. An LLM could help with avoiding duplicate questions. It could analyze the content of the question and point out stuff like missing logs before the question is submitted to guide beginners through the basics.
If someone manages to turn an LLM into a good context-aware search engine, that would also make sense for SO.
But somehow no one seems to actually use LLMs given how much bullshit they talk about them.
If you want answers from a computer, you can get them from chat GPT. The best SO is going to do with that is spend a lot of money to repackage it into a feature people probably weren’t looking for on SO to begin with.
I agree there are other indirect applications of ML models than just generating answers. And I hope ML can help to soften some of the “edges” in UX. A lot of that can be done with BERT or even more primitive statistical methods though.
Please god no. Why this tendency to grow something you like until it includes everything?
There are other places to discuss including the actual Reddit. SO is about question-answers, that's why it got popular. You try to turn it into Reddit, you will add Reddit-scale moderation overhead across 100000 simultaneously running threads to already existing moderation overhead for actual answers. If you didn't notice Reddit can't even manage their own overhead so they freeze discussions after a while.
This is the opposite of what SO should do, which is focus on discovering and improving existing information instead of adding more and more ways to contribute low-effort junk
For new users the edit does have to get approved by 3 other users to give some check on vandalisim.
Thus answers are wikis.
This exists in the form of chat, and extended comment discussions migrate there. That being said, any question that requires extensive discussion is likely not a very good fit for SO because it's probably ambiguous, unclear or otherwise hard to answer exactly.
Or move it to zulip or zulip like so it's at least threaded in a way that works over months. But as is it's just two different purpose tool and cramming one's use case in the other just doesn't fit.
For one, it's almost never a chat. You go there, and the other person who might be talking last posted 12 hours ago. That's not an immediate conversation. These are still "comments". It's a clumsy interface. Then, despite no longer having to be a "comment on a question" like in the main interface, it's still this crappy pseudo-IRC that doesn't give you full markdown to work with.
> That being said, any question that requires extensive discussion is likely not a very good fit for SO
Such they claim.
So basically SO isn't a tool to improve your skills but one to turn you into a copy/paste monkey?
Wikipedia isn't run like a business, but relies on charity. Wikipedia doesn't have to be profitable to exist so of course it isn't run like Stack Overflow, Wikipedia is backed by a foundation that gets plenty of donations. Maybe SO should adopt that model, but that's an entire different question.
If Wikipedia had to be profitable it would be a very different platform.
Both are crowdsourced (so are reddit, facebook and youtube), and the similarities stop there.
I use GPT to implement classes with many interfaces. Even though I often have to make corrections, it's still way faster than looking up the documentation for each of these interfaces. Saves a lot of time in these cases, all the more so I don't have to ponder on which interface in the class hierarchy tree I need to implement.
It could use it to improve its search. However, it should be trained on SO, because the language is quite specialized.
A bimbo is a woman who makes up for a lack of intelligence or competence by having sex appeal.
Don't take my word for it: https://www.merriam-webster.com/dictionary/bimbo
Did you mean to use a different word?
Imagine sanity checking old answers for new versions of language or Library. Adding probabilistic merit to new user answers. Asking the question asker to review machine generated answers in troubleshooting questions with little traction.
Obviously, enshittiffication is at work in how these social0-esque sites try to use tech.
I wonder why didn't Wikimedia have a stackoverflow-like site before stackoverflow.
Reading through the joelonsoftware blog was kind of eyeopening, since I didn't know he was one of its creators.
Mods are the worst. Community moderators are ruining SO faster than the CEO.
https://stackoverflow.blog/2018/04/26/stack-overflow-isnt-ve...
What do you do with the fact that your core contributors are toxic manchildren who love the ability to push buttons at people for daring to ask a question? Build something new where they’re not the core contributors.
I did use GPT4 for a lot of the minor things that i am too lazy to remember and would have „relearned“ on SO every time… maybe others have a similar experience and management just sees visits dropping? Maybe they just focus on developing a target audience for a tailored product.
Google algorithm changes have absolutely curbstomped SO visibility, probably due to third-party sites mirroring their content and making it appear to be a linkfarm to google.
It's funny how even as tech users, google is "the internet" and if something disappears off google, it's gone. Even just pushing a site down a ways in results will massively reduce the amount of traffic, even for us technical users.
There have been a lot of "monkeypatches" from users working around it, like adding "reddit" or site:reddit.com or other hints, but fundamentally google result quality has been significantly declining for 10+ years now and it's getting to the point where it no longer surfaces desirable content anymore.
The company was finished in 2020 when they raised $85M in a Series E. The fall after that is inevitable. Even the $40M they took in 2015 is a questionable decision for the very reasons you detail. What did they need tens of millions in investments for?
Raising at too high a valuation is not a position you want to be in. Unless you’re planning to cash in sooner than the ship will start burning.
https://law.gmnz.xyz/2023/06/08/stackoverflow-keycap-reward....
A company needing that many rounds deserves to die. Sorry to have known you SO.
Finding work there was a dream, and while there weren't many roles available, they were usually great quality roles. Finding candidates was also great, and all of them were solid candidates that met the basic bar of "can you write some basic code in a language of your choosing".
To be honest, it felt like SO went downhill when Jeff Atwood left.
If anything, pretending to be completely knowledgeable "since one's a senior" is quite junior behavior.
Screw them for closing that.
Sounds like you could do with learning something new!
My 30+ years experience doesn't matter for shit when I say... want to fix an annoyance in an open source tool using an unfamiliar language. I'm back to "how do you deserialise json" level query. Offical docs are typically either useless auto-generated placeholders or over-detailed rabbit warren not to mention there are usually five ways to do anything and I need to know the blessed approach not just any approach. I want a few lines of sample code and some confidence that it's the approapriate method and not 10 years out of date. It's what a QA site should excel at and exceed ChatRoulleteGPT answers given social proof from real people.
The good of the community and the well-being of the users are completely irrelevant in this strategy.
These stories are common in headlines because it’s newsworthy, not because it’s common in general.
Ex : Microsoft (evil), Oracle (not innovating and evil), IBM (not innovating), Apple (evil)
Being a startup is seen as "good" by many people because you're seen as "trying", if you stay long and raise your prices, you're seen as "milking".
Microsoft is creating a lot of innovation by trying and failing (Windows phone) or succeeding (Office 365) but many people here see them as evil because they ask for money.
Individual McDonald’s Restaurants are barely profitable, most stores bringing in less than $100k profit per year. For how widespread they are, the restaurants themselves really aren’t a great business to be in.
Friedman's idea that corporations that maximising shareholder value has clearly been the single most destructive ideal that has plagued this planet for the past half century, but to too many people in power it is still the rule.
It's also the reason why we still keep burning coal and oil, despite knowing that we're destroying our planet that way.
Nor is the well-being of the company's employees, or at least the stability of their job and the foreseeability of their future.
It is one more example of a situation where a worker's union fighting for the interest of both the workers and the company (in the sense of the platform/product, not the shareholders) would be to the benefits for the longevity of the business and for the users too.
Anan 7: I'm glad you approve.
Spock: I do *not* approve. I understand.
[0]: https://memory-alpha.fandom.com/wiki/A_Taste_of_Armageddon_(...
If you are insulated from what your company does, or constituents do, you can't effectively wield the power granted to you.
Which leads me to an even more unpopular "authoritarian" opinion: the richest people are the furthest out of touch and thus least qualified to wield so much power. There should be a cap on personal wealth. And generational wealth should be prevented from creating dynasties. There should be no billionaires.
So it seems that any solution would require having executive leadership not driven by growth+profit, or removing the disconnect between "quality" and growth+profit.
Bottom up approaches are the only things that will solve these things, so yes you and me have to put our money and talents where our mouths are.
This board tends towards believing in democracy yet we spend half+ of our waking hours existing within non-democratic institutions…they’re not even representative republics, the thing we pass off as pro-democracy today.
If we want it, we have to build it, we have to build safe-guards, and we have to stand by and not loot as much money as possible when the time presents.
We have so far shown we are collectively not capable of doing these things.
Are there any multi-billion dollar organizations that work on the principles we so desire in our own government? Maybe democracy is just a joke and a sentiment for feel-good buy-in from the masses.
“Tech” has devolved into crony- capitalism on crack. But we’ve minted quite a few average-joe millionaires, so yay?
Maybe try this rule in your own company first. If it’s an improvement over the current system you’ll have a competitive advantage. If it doesn’t work, then you wouldn’t have been benevolent.
> the richest people are the furthest out of touch and thus least qualified to wield so much power
I’m not connected to anyone above “vacation-house rich”, but at their level a lot of the power they wield is in experience, communication and connections. If I’m going to invest in a project, I want someone who’s experienced in the domain, can clearly communicate their vision and has connections in the industry. Wealthy people often tick those boxes.
Here are some alternative free market idea for you:
1. Grant all FTEs some type of ownership in the company (stock, options or profit sharing), and do so on a recurring basis as well
2. Peg the CEOs total comp to be a max of 20x the lowest earner in the company. If the CEO gets a bonus, all FTEs get a minimum of 1/20th that amount as well.
Especially the 2nd one. Linking CEO compensation to FTE pay feels like a good way to remind the powerful that their power derives from the consent of their employees. Plus I'm very against the massive inequality that has skyrocketed in the past few decades.
I just wish there was a way to get the powerful to understand that it's not a one way relationship. The 2 parts, FTEs and executives, both rely on each other to function. But so many people don't want that balance, one way or another.
Everyone needs to really learn what the Toyota Production system has taught manufacturing for decades.
This just means, “people shouldn’t be able to own large companies”. You’re going to need strong evidence that relinquishing ownership of any successful company to the government (or whoever you give the absconded shares to) is a way to run a sustainable company.
What large country (50m+) has successfully eliminated billionaires without eliminating most of their own economy?
Why the government? Why not collective ownership by workers, and maybe partly to users in applicable cases?
Also, why would the existence of such big companies be an absolute necessity in the first place?
That's the place which SO can take in a world with AI assistants.
SO should not use AI to generate content, but to organize it and make it searchable.
It could shift towards the analysis of existing and generation of new content when AI is really ready for it, but I don't see it happen within the next couple of years. And at that point a LLM would probably be smart enough as to not have its user require the use of SO.
Maybe limiting the posting of commented AI results to the top 5% or so, because from the content I've seen new users posting (specially the questions), the quality has degraded strongly over the last 10 years. There's close to zero effort in crafting good questions from many of them.
The “work together with AI” is just a transitional phrase. The endgame is to completely replace the employee.
Slightly OT, but it's counter-productive to start a blog / an article with a well known quote that's been used and re-used ad nauseam. It makes the reader suspicious that they're not going to read something truly original, and tells them the author may have a thing for authority.
I think it's best not to use quotes at all, but if you must, find quotes that are 1/ new and 2/ counter-intuitive, or at least funny.
The above quote, in addition to being famous, isn't counter-intuitive; it's an interesting reformulation of a simple observation: of course people don't usually saw off the branch they're sitting on, even if it means ignoring some facts.
Tell me something I don't know.
Then it's a management problem: what were those people hired to do, given that the content is generated and curated by community alone?
Given that the software running StackOverflow has been working for many years without signifiant changes, they could probably have stuck to ten times less employees and generate substantial profit, but eh that doesn't make a good exit valuation…
> Easily $1-200m/year.
What is this supposed to mean?
LOL. "All your free hard work answering questions powered our profits thus far, but y'know it's expensive to keep y'all in check, and now we need to make megabucks, so we're going to replace you with machines for the next stage of our journey."
Somehow adding LLMs to the side of the answers feels incorrect to me in this particular case.
1. https://www.joelonsoftware.com/2008/09/15/stack-overflow-lau... 2. https://www.joelonsoftware.com/2010/02/14/raising-money-for-... 3. https://stackoverflow.blog/2010/05/04/announcing-our-series-... 4. https://web.archive.org/web/20100416154936/http://blog.stack...
With stack overflow it's very odd how chasing some flavor of the year is seen as preferable than expanding and enhancing the community aspect of stack overflow.
Reddit and the ongoing SO-vs-moderator conflicts show precisely that. I stopped using Reddit and mostly stopped using SO, but now what? There's no good alternative. I just help fewer people now.
If you strongly believe that you have a viable long term plan to switch to a profitable model when the time is right, sure. When AI comes along and wrecks your business before you manage to extract the profit then you can say (with hindsight) you made a mistake by waiting too long. I'm not saying this is the case for SO, but it might be. Time will tell.
Getting to profit quickly is a hedge against change. Nothing guarantees that your business will be around forever. If you don't take the money when its there you might lose the chance.
They're investing in teams and knowledge management. Sounds like they're going after the Enterprise.
Stack Overflow absolutely _has_ to integrate AI if there is any chance of it continuing as a viable business. How do we know this? For starters, a massive portion of users have already stopped asking questions on Stack Overflow and switched to asking ChatGPT.
The situation with the moderators and politics and community interaction might have been botched. Maybe. But I suspect that there might not have been any way to shift direction without making the community extremely angry.
It's not human equivalent. But it can provide useful info and is getting better at a rapid pace. They did not have a choice about integrating AI deeply.
Part of this is people failing to understand the capabilities of the leading edge AI models and how that changes things.
most tech companies lay offs have been around 10% mark - and it started before general public's orgasmic-adoption of A.I.
like all platforms, Stack Overfloow, overflowed their market. layoffs and restructuring is a way to address that.
if you account the rise of "tech streaming" platforms like youtube, twitch, and elearning orgs like udemy etc, it's no surprise that students & professionals choose alternatives to Stack.
This was the original problem that Stackoverflow was supposed to solve, and did do at the beginning.
and AFAIK Discord is not indexed by Google, so the answer is likely never to be found. This goes completely against the idea of the knowledge re-use
The author wants a pur8st approach, i think it having a purist approach doesn't work because AI fills some gaps.
This would save a lot of experienced users time
This would make it easier to find the good answers.
Yes, because if there's one thing Stack Overflow needed, it was more negativity. /s
