Anyone else seeing a drastic drop in community activity for Support forums?

The problem is that Google is killing the internet alongside AI taking over.

I think a lot of communities are seeing the same thing, and I don’t necessarily think it’s because forums have suddenly become less valuable.

The biggest change, in my opinion, is that Google is fundamentally changing how people discover information. AI Overviews are answering questions directly in search results, so far fewer people are clicking through to websites than they were even a couple of years ago. At the same time, ChatGPT and other AI tools have become many people’s first stop instead of Google. If someone has a technical question, they’ll often ask an AI before they ever think about searching for a forum now. ( Why wait fopr an answer when a lot of the time you can get it instantly)

That means forums can’t rely on search traffic like they used to.

One thing I’ve noticed is how much my own habits have changed since using this Discourse forum.

A few years ago, if I needed help I’d post a question here and, more often than not, someone would reply pretty quickly. The support was excellent. If it was something particularly complex, the response was often along the lines of “post it on the Marketplace,” which was completely understandable.

These days, though, if I run into a technical issue, I go to ChatGPT first. Around 90% of the time it gives me the solution within a couple of minutes. It’s even written custom plugins and components for me that work brilliantly.

I think that’s a wider shift we’re seeing. Once people get past the initial learning curve of Discourse—or really any platform—the support forum naturally becomes less relevant because AI can often solve technical problems faster than waiting for someone to reply.

Perhaps the answers were indeed accurate, or you were able to verify the output to ensure it was true/reliable/safe.

My experience of using AI, and becoming warier of their chaotic nature day after day, show me that they are way less trustable I previously thought. And every single day I carefully use AI to get information or answers, I notice mistakes, unreliable or simply false information. The more prompts you send (the longer the conversation is), the more it happens. But even single prompts often returns untrustworthy info.

Of course, it also heavily depends on what I’m looking for. I agree that ask.discourse or general AI can return very accurate info about Discourse, other software, command-line tools, etc.
I ask it to return a FFmpeg command, it’ll do what I ask most of the time.
I ask a developer AI agent to make a full Discourse plugin, I’ll check security features twice more than once!
I ask for accurate information about animals, biology stuff, diseases, medicine, whatever has information scattered among many websites and research papers, it’ll very quickly hallucinate information, quotes, will miss-source, or even lie, if I can use this word. To the point I can absolutely not rely on the output. The answers should be considered nothing more than hints, pointers, that we can use to look by ourself for actual data.

Now, I’ve, like many, answered “try asking ask.discourse.com” to some users here on meta. . What can I say! That robot just works^[1]!

1. well, most of the time ↩︎

My own “User Visits” report is very flat. (I run a hobby forum, maybe that’s different in character)

I see there’s a breakdown Top Traffic Sources (“External sources that have linked to this site the most.”) which shows my traffic over the past 30 days comes mostly from Google, then Duckduckgo, etc etc, with ChapGPT at 9th place, and well under 1% of both Clicks and Topics. In January’s stats it didn’t even make the 10th place.

(Maybe one could query these traffic sources as a time series? Would that be a handy report?)

I’m seeing this more and more myself. I think artificial intelligence is such a misnomer because it’s just not correct often enough. All we have at the moment is artificial information - and lots of it!

This is an interesting and wide spread problem. I’ve seen the traffic of a lot of websites pretty much halve in the last 6-12 months. In the EU there is talk about google being held accountable for the fidelity of the summaries. One can only hope.

In an attempt to get back on topic, I like the coffee shop analogy, and I’m really curious how we’ll look back at this situation in 5 years. I think online communities had a bit of a boom in the last 10 years and we didn’t really appreciate it. I think they’ll become more underground again for a while. Should that occur, I hope it results in more human participants rather than bots(due to lower financial value of the traffic). Those human heavy communities may end up building an inherent value like they did back in the early internet days. Then we get another golden age (hopefully).

There is no denying the power of AI. To ignore it and not use it is to be foolish.

This! I recently had a programming task. When it comes to computer programming AI is like someone hiring you a really bright assistant to help you… you just have to remember that your new assistant might be high… or completely lost their mind

I’ve been a hobbyist programmer my entire adult life. I’ll often choose the language I program in based on what will work best or be most well suited for the task even if I don’t know the language. Recently Google in AI mode basically taught me Python. In about 3 weeks I went from having copy pasted a few python scripts here and there to writing an entire pipeline. Could I have done it without AI? Yes but it would have taken months instead of weeks.

That’s hard to ignore

Not so. One might say that if one took a very utilitarian view. But there are other ways of thinking. You surely know in your heart that this is a controversial position you’re taking. So you can infer that some (many) people feel otherwise. To deny the validity of their feelings, perhaps that is foolish?

I’m not sure what I said that might lead you to think I am denying the validity of other’s feelings about AI. I certainly do not deny the validity of anyone’s feeling on the matter and in fact probably share many of the apprehensions others have toward the subject.

I apologize because I did not realize what I stated was controversial and in fact had I known others would find it so, I probably would not have publicly stated my opinion.

This I will plead guilty of, I can’t help it. I am an engineer and take a very utilitarian view of everything. In fact it may be that my very utilitarian view of life in general may be what caused me to not realize what I was saying would be considered controversial by others. I sincerely apologize if I offended anyone and I will be happy to remove my opinion from my previous post if what I said others find offensive.

And this right here is another reason why forums/social media interaction is dying. Andrew who cares if you offended someone by your opinion. That’s your opinion and your entitled to it.

Let me walk back a bit - I was triggered. Attitudes to AI are strongly polarised, it’s not clear that both sides have a good understanding of the other perspective, and it feels to me that our best future is where both sides do try to understand the other. Any kind of dismissing dissent as “it’s just a tool” gets to me, as would “it’s evil and must be destroyed” which is equally reductive. Not saying you said either, but that’s where I’m coming from. It’s not about my feelings being hurt, it’s about the polarisation.

oh, thank you for that, I think I understand your comment much better now. I agree, it does seem to polarize people’s opinion and extremes are rarely good.

I was probably being somewhat flippant with my use of words and I’ll admit it was meant to be mildly amusing but instead showed ignorance to others opinions

I’m going to be honest, sincere, and perhaps a little pessimistic skeptical because just as most developers, engineers and scientists —brilliant minds in their own fields— don’t understand the true value of encryption today, they also don’t understand where all of this is headed.

The goal is to isolate because that makes it easier to inject the dopamine effect and make more money. The future in 10 years for the opposite kind of interaction is dark (hidden, niche, difficult, not mainstream at all).

You know, try to host your files or share your music without a subscription payment nowdays. The feature hides the training to control, filter and replace.

Those who own internet trained and are training the solution with our human interactions and they are the same who recieves the monthly API payment from everyone who ‘chose’ to work for them.

We are in the transition. IRC is a thing again like self-hosted media servers and sharing via P2P. Felt like a resurrection but it is like Casettes was a thing when we saw the firsts upcoming CDs.

On a ~5-10 older Internet where everyone needs to be identified from a KYCed identity, there is no place to be free again. When the general consensus will be to complain, the only road will be to left the controlled and dominated clearnet.

Social media interaction is gone: IG reels, YT shorts, X ex-tweets are the rule, seconds of attention and then moving on. Infinite scroll. No focus, no emotions, no discussion, no time, no attention at all. Reddit threads are 70% AI garbage. Forums are literally dying, if they are not full of bots.

Perhaps in the data era where all our life is into, the main fight about power dynamics is not on the battlefield anymore but in keeping the value in the basics and simple things.

stallman1080×1826 239 KB

The Future of the Internet — And How to Stop It (Zittrain, 2008)

This is why i’m building our new site around this, as long as we retain traffic that’s a massive plus IMO. People will scroll and read a paragraph

I’m on the same road from different perspectives/tools and really like what you are building on. Please keep going!

A simple note and real fact to be aware building healthy communities today is preventing that some of keep engagement motivation force us to the other side.

You know, so much social or IA features and then the discussions are gone. So much features to implement and then we loss the momentum, focus or point.

Forums are valuable from the opposite philosophy: creativity, shared human knowledge, time, deep focus, purpose.

As long as we can browse the internet without using fingerprint or facial recognition, I understand that the middle path is always the top choice

What a lot of forum owners don’t realise also is that the Wordpress/blog game in a lot of niches is even deader. Google is now prioritising forums/human thoughts in the search results. I went from 200k unique visitors a month over 4 websites to 2k uniques. Now that still hurts to this day.

I have found the stats to be a guide, between bots that don’t comply to standards and spammers the numbers have been inflated. I have aggressively blocked bots, and suspect user-agents and rate limited scrapers, enabling geo-blocking knocked about 30% off spam registrations. Valid logged in traffic has been quite flat. We prohibit AI generated or assisted posts, on a very technical forum it is unhelpful and creates a terrible written tone that does not encourage engagement. I have, ironically, used AI assisted automation to score topics for common AI hallmarks and if multiple are found it is auto-flagged. So far it seems to be about 95% accurate. There have been numerous efforts to create a technical model that works for the subject matter but it is incorrect enough to be a terrible idea. Too many new users are relying on one of the major llm’s as their primary source of information, and they usually find their way to us once they have made an expensive mistake. AI is a powerful tool in skilled hands. In unskilled and inexperienced hands it is a disaster that flatters and rewards incompetence.

Maybe 3 or 4 years ago I decided to explore phpBB, because as a musician who got tired of social media, I wanted to build a community not only to share my music, but also to share what I’ve been learning over 30+ years of making music. At the same time, unlike social media, I wanted to build a place where everyone could share their own content. No algorithms. At the time I remember telling my brother that forums and groups were the future of “social media”, because people are getting tired of the meaningless of social media platforms. I still think the same way about it. I still believe communities and groups are the next thing, if done properly.

Going back what @Helga_Razinkova originally posted, my first thought was “the decline is because of AI”. It was not even something I had to think much about it. I use it a lot myself, just because it’s fast. And when the answer the LLM provides actually solves a problem, there’s really not much we can “agree or disagree” with. It’s there. Does it always give good answers? No, but 99% of the time, depending on what is being asked and how it’s being asked, I see my questions being answered pretty accurately. And when something seems a bit off, the good old Google helps as a good complement. So yeah, AI is definitely making people go to forums/communities/groups less and less frequently. And that leads me to the next step community owners should be focusing on:

Don’t just use your communities as a place to answer questions an LLM can easily answer. Bring other value so people can engage with it. Let me use an analogy: we like having friends, not because they are there to help us when we need them. We like to go out for dinner, a walk, even to listen to their problems and provide our own solutions, right? The issue with communities is that most of them are that “one way” platform. “You ask me, the owner of product XYZ, and I answer”. Well, when the LLM can provide the same answer, in 10 seconds, there’s nothing more a community can provide (or at least, not as much).

We need to bring other sources of value. That’s one of the reasons I’ve been writing articles and more articles I want to share on my community, once I finally put it out there. I don’t want people to just come and ask questions. I want to provide the content, like a social media platform, that people can learn from, challenge me, etc. I want the users to start real conversations, not just based on issues an LLM can answer, but because there’s genuine interest in talking to other people. That’s what a real community is. Otherwise, it’s just a customer support channel, that eventually any LLM can replace.

I understand that not all communities have much to share (or so we think…), other than customer support, but if we dig deeper, maybe we do…?

Just my 2 cents…

The challenge is all great content gets scraped, assimilated into the next model, or is delivered direct to an AI user in their chat box, presented as it’s own content, bypassing the creator and their site. It’s a catch 22. AI typically does not drive traffic as conventional search used to do. Adapting to this new reality isn’t easy. I have no way to measure how effective the LLMS.txt really is, and that is only if the scrapers honour standards.

Declines are driven by changing behaviour and expectations. AI can fake engagement but it cannot offer a breadth of specialist knowledge and different perspectives based on real life experience. It is the human element that will keep people coming back.

AI cannot offer a sense of community.

But think about this: LLMs can only copy what has been created. It cannot fully predict what I’m posting next. I can create an “average” estimate based on the past, but it can’t fully know what I’m writing next. I don’t really care much if it gets scrapped and all that. I’ve been benefiting from all the content it scrapped from others, right? I don’t think the solution is to go against it, but how to use it, and especially, how not to get carried away by what it “might do”. we have two options: post stuff online, or keep all to ourselves, preventing others from enjoying what we have to share. And what an empty life if we all decide we will not share anything, “because AI…”

Exactly. That’s why I’m saying that if forums keep doing the same thing, and people expect different results, the decline is to be expected. For example, if I have a favorite artist and he has a community where he engages often, I don’t even want to be in a place like Instagram where he also posts, where I know a lot of the comments are bots. At least in a community, we can more easily control that, especially because in a community we don’t have 2 billion users (plus the millions of bots) all looking for the same attention.

I agree. But again, if the community only offers answers to questions that an LLM can answer, do we really have a community?

In my hobby community, I’m hoping most people are coming to get answers to questions like

• what have people been doing?
• what have they found, or made, or fixed? Or broken, or lost? Or learned, or read, or bought?

This works so long as a fraction of people pre-emptively post the answers to those questions.

These aren’t, I think, questions that a chatbot can answer.