It's kind of... hilarious that Reddit has managed to wait so long that they've ended up shooting for an IPO just after the company's third largest individual shareholder (Altman) created an existential threat to destroy reddit.
A couple of years ago I think you could've made a decent argument for reddit, it's never going to have as good economics as Facebook but there's definitely value and communities there. It's a bit like twitter- big reach but poor economics.
The problem is that with Generative AI there's just an enormous question mark over the business. What is Reddit going to do to ensure that their pseudonymous user base are real people and it isn't just going to be over-run with bots. They already had a smaller version of this problem - astroturf advertising, but now that problem is on steriods. And because they don't have any money they don't have the resources to solve this problem!
If you want to buy this stock I just don't know how you justify to yourself that any of the numbers are going to be real. How many unique users are real users? How many ads that were displayed appeared in front of real eyes? How many comments are actually hand written? And if you as a shareholder don't know, the hell is Reddit going to persuade advertisers?
I think what you're pointing out is systematic risk for all social media or companies driven by online advertising. You can think Reddit is a good play relative to those companies. Just buy reddit and short snap for instance
For instance, reddit has these things going for it:
Data - sure bots will influence Reddit but there are still a lot of users today. Sophisticated anti bot technology could help keep that somewhat down. Besides if gen ai becomes so good such that it can generate high signal training data for itself, presumably the bots could benefit Reddit. Sometimes I just want information or discussion, and I don't care if I'm talking to a dog on the other end
Low RPU relative to competitors - the demon is different than Facebook or Instagram but the rev per user is absurdly low and can probably grow somewhat
What twitter pulled off - twitter ditched something like 80% of it's staff and is pushing out product faster than ever. Think about how much it changed in the last two years. Editable comments, subscriptions, revenue sharing, blue checkmark for sale, alternative check marks for govts and org, culling of old accounts, API restrictions, forcing people to put parody in acct, xAI, long form video, prob more things I'm forgetting. If the right person decides to cut out all the bullshit and prioritize things like making the website not a flaming dumpster fire, no reason that couldn't dramatically impact usage
It's uniquely bad for Reddit for a few reasons - more pseudonymous users, less 1:1 direct communication, and far more text based than competitors like Insta.
Twitter is also a good point - Reddit looked a hell of a lot more valuable when it was sitting next to a $44Bn Twitter. Now it's sitting next to a steaming pile of far-right rubble that might not even have equity value.
>If the right person decides to cut out all the bullshit and prioritize things
How can that be an argument in favor? If the right management team magically came around, you could also buy Boeing based on that hope. But I don't think any sane investor would do that.
Just start firing people. Meta learned the same thing and cut something like 25% of the workforce. It's organizational debt, start removing layers. Because so many of the people are doing destructive make work things, throwing darts to lay off entire teams would even prob even be net positive, but ideally it would be more thoughtful.
There's no reason the app sucks so bad. Seems like this would be a priority?
Snap is an odd example to bring up, since they seem the most well guarded against becoming dominated by bots. It's primarily a private messaging platform, with a couple of broadcast social media features. I'd assume the vast majority of Snapchat's traffic is between people who have met each other in real life.
How many real users are on HN? It was pretty obvious years ago that some users were AI bots. The bots used fake words and sounded like it was full of academia jargon but the words were not real words.
After AI advances in the past couple years I am almost to the conclusion that I will stop posting. Especially considering the close ties with HN and Altman, I am not here to fine tune someone’s AI.
The theories about the origins of these commenters were amazing. There's nothing to say that other people are not doing the exact same thing right now here on HN and other places. And very few people would even notice, because open LLMs have progressed significantly since then.
I was perplexed trying to read few comments and understand what they were saying. As soon as a user said these look like they were generated using markov chains I knew something was up.
If I can find the comment I’ll share it.
Here is one example[1]. Trying to find a better one.
How do you know that you aren't a bot? Maybe what you feel are your own genuine experiences of the world is just part of the simulation. Now and again they spin you up to post on HN, feeding a new set of memories that have elapsed since your last output.
I am a bot, by the way. Altman made me post this to sniff out competitor bots.
Wait, what? Aren't all of the things that Altman is doing now making all of the data that Reddit is sitting on (and its capacity to generate new ones) EVEN MORE valuable? Didn't they just announce some data sharing deal to train an LLM worth many millions not too long ago?
So this is the story that places like Reddit and Quora would like to tell. They'd like to say that they have this treasure trove of data which can be used to train the models of the future, but that doesn't seem to be what most experts actually beleive.
Firstly, Reddit is closing the stable door after the horse has bolted- they were open enough for long enough for most serious players to have most of that data.
Second, LLMs are poisoning the well. Reddit's content isn't going to be good training data if it is itself the output of LLMs and that is going to happen.
But finally and most importantly, it's not good! The open internet isn't great to train LLMs. A really important basic element of creating good models is training them on good data, and data from reddit as a source is terrible! Reddit as a source is about on par with just scraping the open web. If training data does become valuable you should be looking at places like the New York Times or scientific journals - places that have large volumes of quality training data.
While I am sure an LLM only trained on publications would be great, I can also picture it being the most annoyingly square AI to interact with. Just generally being painfully out of touch with how average humans speak and interact.
I guess the ideal would be training with some kind of source discriminator or internal source quality parser that allows an LLM to be the chill funny guy who can still school your nerdy ass in distributed network architecture.
Sincere question: To the extent that "the open internet" IS valuable data, isn't Reddit in the best possible position to capitalize on it? The mechanic of the website allows for people all over the world to contribute free-form content and then has mechanics built in where the content provided is graded by free human reviewers and moderators, and graded accordingly. Not saying it is perfect data, but surely there is not better alternative to it, no? I just don't understand how this data is "terrible" -- other than very limited things like newspapers or other very highly curated (and limited sources), what other large scale data repos are better?
I’m in a weird position of both valuing Reddit highly as a source of unique and niche information you literally can’t find any where else, while at the same time feeling that the general quality of discourse in the more mainstream subreddits has taken a nose dive and is now pretty much worthless.
The problem is the grading is trash. In some sub-reddits the grading is literally inversely proportional to the quality of the content. In others its inversely proportional to traditional grammar rules. In some places it's highly correlated with misogyny. Like how often do you want your LLM to slowly transition it's response to a question into the Fresh Prince of Bel Air meme?[1]
Compare it to other forms of content - news sources and journals which are extremely well curated, digitized books - a massive back catalogue of well edited and well structured data, Wikipedia - extremely high quality.
I would put it this way, training an LLM on reddit is like teaching a child to play piano entirely by exposing them to freeform jazz.
Wait, aren’t there enough scraped dumps of Reddit data already?
Sure, it is not as convenient as an API, and you may not be able to put Reddit’s logo next to the results, but why would LLMs care about that as long as they can train on the data in some form?
LLMs replace the command line and search. It doesnt need to return AI written content, although summaries are one of the most useful returns. The AI part can just be interpreting what was asked of it.
That said, if I can say "approximate a conversation between two film nerds" and the AI can summarize 100 film reviews on metacritic/rotten tomatoes and then convert them into a coherent dialog, with each party taking specific positions, the AI can replace the middle man reddit user that is regurgitating something they read.
"Most people are other people. Their thoughts are someone else's opinions, their lives a mimicry, their passions a quotation." - Oscar Wilde.
CEO pay for 2023: 341,346 (cash salary) + 98,332,716 (stock) + 93,776,049 (options) + 792,000 (cash bonus)
It's hard to really value the 190M of stock+options, though it seems like it'll be enought to buy some really nice avocado toast.
Anyway though, shouldn't we be celebrating this? Isn't it the hacker news dream, creating a startup that creates zero real physical products, contributes to climate change, drives youths to addiction and worse, all while becoming filthy rich along the way.
yes, we should be celebrating this. also, 1) cash comp was $1m and the $190m equity grant vests over 10 years, 2) see the comment posters username - are we surprised by this take? lol
Reddit seems to have well and truly jumped the shark. Most of their communities are barely moderated wastelands. Bot activity is absolutely rampant. It had a golden age, but I have no idea how this survives in the post LLM world.
Barely moderated?
Part of the issue with modern Reddit is newbies can’t contribute in subreddits without breaking the numerous Byzantine rules powermods make.
63 comments
[ 4.2 ms ] story [ 124 ms ] threadbarnum was right
A couple of years ago I think you could've made a decent argument for reddit, it's never going to have as good economics as Facebook but there's definitely value and communities there. It's a bit like twitter- big reach but poor economics.
The problem is that with Generative AI there's just an enormous question mark over the business. What is Reddit going to do to ensure that their pseudonymous user base are real people and it isn't just going to be over-run with bots. They already had a smaller version of this problem - astroturf advertising, but now that problem is on steriods. And because they don't have any money they don't have the resources to solve this problem!
If you want to buy this stock I just don't know how you justify to yourself that any of the numbers are going to be real. How many unique users are real users? How many ads that were displayed appeared in front of real eyes? How many comments are actually hand written? And if you as a shareholder don't know, the hell is Reddit going to persuade advertisers?
For instance, reddit has these things going for it:
Data - sure bots will influence Reddit but there are still a lot of users today. Sophisticated anti bot technology could help keep that somewhat down. Besides if gen ai becomes so good such that it can generate high signal training data for itself, presumably the bots could benefit Reddit. Sometimes I just want information or discussion, and I don't care if I'm talking to a dog on the other end
Low RPU relative to competitors - the demon is different than Facebook or Instagram but the rev per user is absurdly low and can probably grow somewhat
What twitter pulled off - twitter ditched something like 80% of it's staff and is pushing out product faster than ever. Think about how much it changed in the last two years. Editable comments, subscriptions, revenue sharing, blue checkmark for sale, alternative check marks for govts and org, culling of old accounts, API restrictions, forcing people to put parody in acct, xAI, long form video, prob more things I'm forgetting. If the right person decides to cut out all the bullshit and prioritize things like making the website not a flaming dumpster fire, no reason that couldn't dramatically impact usage
Twitter is also a good point - Reddit looked a hell of a lot more valuable when it was sitting next to a $44Bn Twitter. Now it's sitting next to a steaming pile of far-right rubble that might not even have equity value.
How can that be an argument in favor? If the right management team magically came around, you could also buy Boeing based on that hope. But I don't think any sane investor would do that.
There's no reason the app sucks so bad. Seems like this would be a priority?
After AI advances in the past couple years I am almost to the conclusion that I will stop posting. Especially considering the close ties with HN and Altman, I am not here to fine tune someone’s AI.
The theories about the origins of these commenters were amazing. There's nothing to say that other people are not doing the exact same thing right now here on HN and other places. And very few people would even notice, because open LLMs have progressed significantly since then.
If I can find the comment I’ll share it.
Here is one example[1]. Trying to find a better one.
[1]https://news.ycombinator.com/item?id=26848661
I am a bot, by the way. Altman made me post this to sniff out competitor bots.
Firstly, Reddit is closing the stable door after the horse has bolted- they were open enough for long enough for most serious players to have most of that data.
Second, LLMs are poisoning the well. Reddit's content isn't going to be good training data if it is itself the output of LLMs and that is going to happen.
But finally and most importantly, it's not good! The open internet isn't great to train LLMs. A really important basic element of creating good models is training them on good data, and data from reddit as a source is terrible! Reddit as a source is about on par with just scraping the open web. If training data does become valuable you should be looking at places like the New York Times or scientific journals - places that have large volumes of quality training data.
I guess the ideal would be training with some kind of source discriminator or internal source quality parser that allows an LLM to be the chill funny guy who can still school your nerdy ass in distributed network architecture.
The first explains the answer, the second translates the answer into layman, and the first validates that the translation is still mostly correct.
Compare it to other forms of content - news sources and journals which are extremely well curated, digitized books - a massive back catalogue of well edited and well structured data, Wikipedia - extremely high quality.
I would put it this way, training an LLM on reddit is like teaching a child to play piano entirely by exposing them to freeform jazz.
[1]: https://knowyourmeme.com/memes/bel-air-fresh-prince
Let's be real: If Reddit is at risk, the whole web is.
If "AI" content becomes widespread and the tooling accessible, the web is dead.
LLMs replace the command line and search. It doesnt need to return AI written content, although summaries are one of the most useful returns. The AI part can just be interpreting what was asked of it.
That said, if I can say "approximate a conversation between two film nerds" and the AI can summarize 100 film reviews on metacritic/rotten tomatoes and then convert them into a coherent dialog, with each party taking specific positions, the AI can replace the middle man reddit user that is regurgitating something they read.
"Most people are other people. Their thoughts are someone else's opinions, their lives a mimicry, their passions a quotation." - Oscar Wilde.
Company exists simply to extract money for the CEO. You'd have to be insane to buy any shares.
CEO pay for 2023: 341,346 (cash salary) + 98,332,716 (stock) + 93,776,049 (options) + 792,000 (cash bonus)
It's hard to really value the 190M of stock+options, though it seems like it'll be enought to buy some really nice avocado toast.
Anyway though, shouldn't we be celebrating this? Isn't it the hacker news dream, creating a startup that creates zero real physical products, contributes to climate change, drives youths to addiction and worse, all while becoming filthy rich along the way.
re (1): https://x.com/eshear/status/1761483016936046596?s=20
But I would hope that we hold public companies to a better standard.
Putting in place a 10 year options scheme before IPO so future public shareholders can't vote it down just smells.
Or someone just buy all the stock and just end it, pretty please.
Quick, someone tell Elon
From there my power fantasy is firing the ceo and reopening the api so we can go back to Apollo.
https://news.ycombinator.com/item?id=39669162
The best predictor that a stock is going to do well is bad financials.
My condolences