And I'm taking over openai
Reddit is taking over Google
261–270 of 349 posts
Re: Reddit is taking over Google
#262Earlier quoted context omitted.
Who are you proposing pays for content moderation then? Reddit clearly can’t afford it
Reddit has demonstrated that its moderation system is not stable and results in garbage. Try some different systems, for example: 1 -- The moderators pay for the privilege of moderating. or 2 -- No moderators at all. Let me decide what I want to see via some kind of interface to an "AI thingie".
Requiring moderators to pay for the right to moderate would backfire. The groups which would pay the most are political parties, foreign governments, PR firms, and corporate surrogates. All, would pretend they are “neutral” but in fact would slant things in their favor. For example, a cars sub-reddit would talk about how Brand X is “awesome” while Brand Y is “terrible”.
Re: Reddit is taking over Google
#263> Google and Reddit announced a blockbuster deal that would let Google train its AI models on Reddit content.
Whenever I land on Reddit for a topic where I have specialised technical knowledge, the comments are at best incomplete but more often plain wrong or uninformed. Yet always written with a tone of authority.
Reddit is definitely not the answer for accurate LLM fact training.
Re: Reddit is taking over Google
#264Earlier quoted context omitted.
It's the last bastion of original content before the AI storm overruns search entirely. And likely puts reddit firmly in the the AI farm crosshairs as well, so reddit will be overrun. I think the only good content will be in archived discord channels, or obscure boards that aren't worth aiming AI farms at, but even then, it will probably get scanned and targeted. The only thing that might work is identifying AI infor…
> intentional misspelling and curse words will start to be used to "scare off" AIs and bots If they deem the content valuable enough, I suspect they'd instead just correct for it (e.g. auto-correct mispellings; replace and/or omit curse words). It may get to the point wherein the only way to secure content from AI ingestion is to hide it behind a paywall or otherwise make it non-public, then forbid via the TOS. There…
Criminals, SPAMers, foreign governments, PR firms, unethical corporations, etc. are not going to be stopped by a terms of service document. This will ignore it and keep on spreading disinformation. If AI helps them do this, they will use AI to do it.
Re: Reddit is taking over Google
#265Earlier quoted context omitted.
> Because Reddit has horrible search Sorry, but I've heard this so often, and I just don't know what people mean. I use the search whenever I need to get something from a specific subreddit, and it just seems fine. Also, why would a major website like Reddit have such a bad search? Aren't mostly fine search algorithms freely available?
One of the big mysteries of the software industry. Not specifically singling out reddit here, since I don't even search on reddit: Why can we not simply have search functions, that at least optionally do a substring search? If I had a $currency-unit for every search that fails at this basic thing, I wouldn't need to work any longer.
Search the same thing rephrased slightly differently 5 times to find the right result, and I can show you 5 sets of ads related closely to the topic of your search.
Re: Reddit is taking over Google
#266Earlier quoted context omitted.
> That was done specifically for SEO purposes. As a human I also like it when I can see what topic a URL points to (eg. when it's being shared on HN, or pasted into source code, etc). URLs that just have an opaque ID or terse filename are not very user-friendly.
The problem is you can't trust it: people can modify that part of the URL arbitrarily and so you think the URL has meaning, but it doesn't.
It depends how it is implemented. After extracting the ID and looking up the post you can then check that the "slug" matches. Return a 404 if it doesn't.
This may add a bit of complexity (if the slug data is mutable you need to allow any past versions or freeze it in creation) but isn't too hard if this is the approach that you want to take.
Re: Reddit is taking over Google
#267Earlier quoted context omitted.
It's mind-boggling to me that Google didn't create a spec to work with special partners whereby they could get syndicated data in a format that was easy to ingest, and easy for their partners to produce. Lighting dollars on fire just to serve up pages to Googlebot, when you could just periodically dump a journal of updates to Google, is just crazy imo. On edit: if only there were some way to do some kind of really si…
That syndicated data format may or may not be up-to-date.
If it’s specifically accessible to special partners, Google could just end their partnership if Reddit is serving up false information to them and go back to the old way of scraping.
Re: Reddit is taking over Google
#268Earlier quoted context omitted.
It's mind-boggling to me that Google didn't create a spec to work with special partners whereby they could get syndicated data in a format that was easy to ingest, and easy for their partners to produce. Lighting dollars on fire just to serve up pages to Googlebot, when you could just periodically dump a journal of updates to Google, is just crazy imo. On edit: if only there were some way to do some kind of really si…
You’d have to trust that the data being dumped was 100% identical to the actual pages users would eventually see, or you could end up with very weird (including dangerous) behavior Of course, I know that some version of this can and does occur with classic web scraping too, but that is an arms race that a search engine can win
You can just spot check a tiny fraction of the data to validate this big it doesn't match the the site gets blocked.
Re: Reddit is taking over Google
#269Earlier quoted context omitted.
It's mind-boggling to me that Google didn't create a spec to work with special partners whereby they could get syndicated data in a format that was easy to ingest, and easy for their partners to produce. Lighting dollars on fire just to serve up pages to Googlebot, when you could just periodically dump a journal of updates to Google, is just crazy imo. On edit: if only there were some way to do some kind of really si…
Isn't that what WeSub (previously PubSubHubbub) is? https://en.wikipedia.org/wiki/WebSub
I don't think Google uses this in a major way. It also probably isn't the best fit for this use case.
1. You can really only subscribe to particular URLs. So this would require millions of subscriptions. It would only make sense if most of your pages are changing every few days.
2. You need to also subscribe to feeds to fine new content.
Re: Reddit is taking over Google
#270To this day, my most public contribution to reddit is that I wrote the code to put the title of the post in the URL. That was done specifically for SEO purposes. It was pretty much the only SEO optimization we ever did (along with a few DOM changes), because shortly after that, Google basically dedicated engineering effort specifically to crawling reddit. So much so that we lost the "crawl rate" button in our SEO adm…
Nowadays the reddit engineers perverted the link system for old reddit users. New reddit users link to posts using a syntax that for old reddit users returns the submit a new post page. Also, just to twist the knife further, old reddit users can't open full resolution images hosted on reddit anymore. Click a link and you get thrown to this url: https://old.reddit.com/r/funny/comments/media/nice_hat/?url=...
I just find it hilarious that old.reddit is one version, new.reddit is another, and www.reddit is yet another.