Yes. In the beginning they didn't ban opinion based posts (that's why you can still find some of them that were left up for "historical value").

I liked Joel on Software, I liked Coding Horror, and I liked the idea that two internet guys could just identify a problem like that, start a company and fix it.

There was a Goldilocks period of several years where contributing answers was fun. I joined in 2010 and was most active until about 2016. It felt good to help people and since it was in the open, it felt like a resume booster as well, like having an active GitHub profile.

> Did anyone actually like StackOverflow?

Absolutely! It answered so many questions I had. I remember when it was relatively new, and I'd just heard about it, it profoundly changed how I found answers to programming questions. Suddenly I could find people having similar problems that I had, who had asked the question, and had actually gotten a useful answer from someone who knew!

It wasn't perfect, of course (nothing is), but it was orders of magnitude better sifting through crap blog posts and confusing reference material. (Not to mention sites like experts-exchange.)

Stack Overflow was a nice experience for me because I was able to hit 2k reputation fairly quickly, in just 30 days of posting and 6 weeks calendar time. That being said, it never had the community feel of places I spent during my formative years, which were more on forums and IRC.

Here's a conference talk I gave on how to gain Stack Overflow reputation from back in 2018, selected out of 5 submitted talks. It's amazing how fast times have changed from before, during, and now after.

https://grokify.github.io/stackoverflow-the-hard-way/

I did originally, when it collected a bunch of obscure knowledge and made it searchable and useful. It was fun and rewarding to put things you knew into the common knowledge pool, and everyone celebrated a successful competitor to Experts Exchange. The SO model had a few major flaws that became impossible to ignore after it was entrenched. First, the reward scheme rewards the exact opposite of what it should incentivise: common questions are hit by many users and therefore attract lots of upvotes while answering the really hard stuff often meant you didn't even get your answer marked as "accepted" (because the OP had given up and stopped checking the site). Second, the site deliberately cultivated an "editor caste" in the Wikipedia style before the failure modes of that model were well-known: well-intentioned newbies get shut down by miserable yet untouchable people who play (and sometimes help write) the site's rules. Third, the stated desire to identify canonical answers to questions had no clear way to handle the evolution of the software people were talking about. So you'd have highly upvoted answers that might have been referencing deprecated libraries, and it was very hard for the newer answer to gain traction via either internal or external search.

It was also unfortunately before the retro boom of the 2020s, so questions about older arcana were often vulnerable to being closed instead of answered.

I had a pretty good time asking a question about Prolog. It was a really interesting experience knowing that there's someone out there that high proficiency in a very niche language, patiently explaining to me an issue that they have probably heard a million times from yet another imperative programmer. They even have their own website advocating for Prolog, etc.

Now, I could imagine an LLM would be able to do the same. However, I understand that this is only possible because of people like them. I don't think the youngins that started with LLM directly would appreciate the humongous amount of data and discussions online that enables that. The internet is so much bigger than just Google, Facebook, Youtube and Twitter.

It was great 15 years ago before the Iron Law of Bureaucracy kicked in and the powermods took over.

You may substitute StackOverflow with Quora and your answer would not get less valid.

I always thought it was a perfectly fine service for lookup. Asking a question though required a very specific confluence of circumstances to actually be a useful thing to do, so I only did that like 1-2 times in many years of reading it from google.

I liked StackOverflow for the answers.

But asking questions was hell.

I liked it, still find good answers, and it was gratifying to provide answers when I could.

StackOverflow was great when I was a very junior dev working on JavaScript apps. Anytime I ran into a roadblock, there was often a relevant post there to help me. As I become more competent though, I realized that reading the documentation directly was usually a much better way to get answers to my questions, and I stopped visiting.

I left it around 2014 and the graph does show that. I think what happened is that around d 2014, it gets so many people from search that it could simply abuse users left and right and still have tons of new questions everyday.

Yes, I liked it. When it debuted it was a massive improvement over expertsexchange, who had previously dominated the Google searches with bait and switch previews.

It may not have aged well but to say it was always crap is rewriting history.

I have found StackOverflow useful on rare occasion, but the friction, idiotic moderation culture, and high noise-to-signal ratio usually made it somewhere I didn't want to return to.

I was just thinking to my self the other day how it's nice I don't need to stop what I'm doing to make a question that's answerable by someone else. Ai can answer my question without me spending time recreating the problem and stripping out all of the irrelevant context

It was pretty dang useful when there was no alternative, and I’m sure that many people physically could not have performed their jobs without copy-pasting from it.

But yeah, I don’t know how anyone could have any affection or nostalgia for it, people were massive jerks and it was not a pleasant place.

I thought StackOverflow was pretty great. This is an unpopular opinion but I think a lot of the questions that were closed really deserved to be closed. Otherwise it would have been a firehose of the same basic questions over and over again. For every person who posted a question and got mad that it was closed, there were probably 100 people who googled something and found a useful StackOverflow answer that was relevant and useful to them although they never posted their own question or even made an account on the site.

For me the main problem of SO isn't even the moderation or human interaction. Even if a question is answered successfully, the entries have a short shelf life because modern APIs move and break so quickly. For example, I tried learning Ansible only through books and SO, and it was just frustrating. ansible_sudo_pass was deprecated for ansible_become_pass, but there are still many books and SO questions that still reference ansible_sudo_pass.

In the Good Old Days (or my rose-tinted memories of them), Java/C books and answers will always work even if it's not idiomatic, and JS/Python material might break once in a decade over a major migration like Python 2 to 3. Now I look at Ansible or Zig, copy a simple toy program from SO or GH, and just find that it doesn't work, because `sudo` became `become` and `fs` became `io`. There is simply no way for books or SO to keep up.

Instagram, YouTube, WhatsApp seemed to work out well.

It’s always more complex than you think

The risk is that beyond hobbyists technology will stagnate. It’s already much more productive using one or two frameworks over the rest of what’s available, and without training data it won’t be able to advance beyond those current popular frameworks.

A lot of companies don't hire at the end of the year because of holidays breaking up schedules, coupled with strapped budgets by that point. New year starts, budgets are refreshed, everyone is back to work so hiring pipelines can roll again. Get hired in early February, on-board for a couple of weeks, really start to dig into work at the end of February and early March. You don't want to look like you don't know anything, so instead of asking your coworker sitting next to you or just a Slack DM away, you throw a question up on Stack Overflow and hope to get an answer that gets you unstuck.

Too early to call? It's hit rock bottom. I've never seen a major site die so completely before.