Live data from Hacker News

Reddit can't build a better search engine

ruky.me

211–220 of 289 posts

Re: Reddit can't build a better search engine

#211
post #92

Earlier quoted context omitted.

Recently have noticed a business creating fictional threads on reddit (multiple on different communities) working as an AD to sell something. Example: „What is the cheapest best cooling for cpu now?” All discussions point to product that more experienced users know its crap. But still the wrong product is at the top with most (botted) upvotes. Companies realized that ppl look for opinions on reddit -> that created de…

Yeah initially I thought Reddit was useful for "give an opinion" type queries, especially where you're getting answers to a newly posted query. Eg my brother managed to get a specialist doctor to give an opinion about our mom's treatment. But for the vast majority of commercially relevant searches, it will get sunk quite soon. It can't be terribly hard to hook up a GPT style bot to harvest karma on various accounts a…

> It can't be terribly hard to hook up a GPT style bot to harvest karma on various accounts

Or more easily, steal and repost images on cute animal subreddits, sometimes adding white borders to the bottom right so duplicate image detection algorithms can't find it.

Re: Reddit can't build a better search engine

#212
post #177

Earlier quoted context omitted.

I think Google is over because the problems we have now are harder and harder to google. That combined with stagnation of constructive and/or simple progress, all caused by lack of energy. Means there is less and less you don't know that Google can teach you. I haven't googled anything important since around 2016. Reddit gives you a narrow subject with people that care about said subject, much much better!

Well there still seems to be a lot of deep content on the web, it's just that google doesn't show you any of it. You may, if you're lucky, get a Wikipedia box. And Wikipedia, for all its breadth, is incredibly shallow on most topics. I submit that the reason you aren't learning anything from google is that it just doesn't surface anything for you to learn from, but usually just the same dozen or so big sites, wikiped…

So why you think this particular website must be the top answer? I looked at googles results and and I disagree, they are not bad, e.g. actual program examples, videos and papers for example.

Re: Reddit can't build a better search engine

#213

Earlier quoted context omitted.

I largely blame Reddit's redesign. Once Reddit became more about infinite scrolling with media content becoming front and center, that made it much easier to gain attention through low-effort gifs and memes. It draws in the lowest common denominator of people looking to scroll through mindless content while on break. More eyeballs and more screen real estate means more opportunities to shill and propagandize. Old Red…

The low effort racist/political/pepe shitpost image macros came across from Facebook. My grandparents love them. I wouldn't mind seeing a lot less of them on Reddit, but that's apparently the only part of the site's userbase that doesn't have adblock installed which is why Reddit caters exclusively to them.

Umm... I guess that can be considered part of it, but somehow I manage to avoid the racist shit without trying.

I'm more broadly talking about how it evolved many of Reddit's communities to be more "look at me!" based as opposed to "what do you think?" or "here's my story". Not every community is like this, but if moderators aren't strict, communities quickly become an amalgam of these:

- Here's me doing a thing loosely related to the topic of the sub! Look and congratulate me!

- Here's a screenshot of some news story with no hyperlink. Get mad with me!

- Here's a meme making fun of breaking news. Laugh and mock with me!

- Here's a product I bought that's loosely related to the topic. Don't I look badass? Discount code in the thread!

- Here I am, [pretending to be] an idiot just getting into the topic. Tell me what to buy!

This is of course an incomplete picture, but this more or less reflects my experience with what I consider to be low quality Reddit content. They are mostly about visual attention and are designed to either promote products, their YouTubes/Instagrams, or make you think that the community believes something, ergo the reader should believe it, too.

Old Reddit wasn't immune to this, but I think there's a definite difference when the UX is designed around passive content consumption versus something more active. The old Reddit interface was ironically more of an active experience because you were forced more to visit individual posts based on the title and thumbnail, meaning posts better be good or fewer people will read them. With the New Reddit's passive content consumption, the visual content is front and center on the screen with no effort by the user other than to keep swiping their finger upwards to move down their feed. This design both invites a different type of user and greater amounts of shilling for products and political memes.

But maybe I'm way off base with how most people use Reddit. I believe I've seen Reddit change overall as well as some individual subs.

To the creddit of Reddit, subs do have the ability to avoid most of these issues. Some of them turn off image posts entirely, which seems to greatly increase the value of a community. Unfortunately, that doesn't stop Reddit as a whole from being a shilling and propaganda machine. People want their dumb pictures.

Re: Reddit can't build a better search engine

#214
I just don't agree with this trope that Google is "bad" or even quantitatively "worse". It's almost always based on:

1. Ads in search results. I don't agree with the premise that ads are necessarily bad. Someone selling a thing may be the best result for a search. As long as ads are labelled as such I'm fine with this;

2. Anecdotes. I can match you anecdote for anecdote;

3. Personalization: it depends on what's included but knowing you're in Chicago changes what's most relevant when yous earch for "auto repair shop"; and

4. Astroturfing, content farms and SEO gaming. To me this is the biggest problem and will be forever an arms race. The article touches on this. It's really a product of views (ie display advertising) and clicks (ie affiliate programs) generating revenue and any search engine will have this problem.

The lesson you should take from using "site:reddit.com" or "reddit" or any other term to Google search like this is evidence of just how hard search is. The 21st century is littered with the corpses of dead "Google killer" search startups.

The article (correctly IMHO) mentions Reddit doesn't get as much attention SEO-wise and that's actually what can make it useful but it misses the main point: there are communities that form around niche topics (eg vacuum cleaners) that are generally resistant to astroturfing.

Re: Reddit can't build a better search engine

#215

Reddit results are not organic at all but maybe appears to be on the surface. Subs have their own "reddit" brands in accordance to the topic. They also have their own "reddit" opinions shared among almost every sub with enough people in it for justify propaganda. I already did my own informal research on this - reddit has transitioned into consumption. I went sub by sub for thousands of posts, even a few years ago, a…

> They also have their own "reddit" opinions shared among almost every sub with enough people in it for justify propaganda.

This has been a problem I've observed when using reddit to choose products. If a bunch of people are going to get together to talk about X, they are going to be way more into X than most people. When I was choosing a mesh network system the conclusion I got from reddit was that spending less than $600-$800 and having a system without a wired backbone would make my internet unusable. It was extremely difficult to find a "good enough for typical users" recommendation since none of the posters were typical users.

Re: Reddit can't build a better search engine

#216
“The best way to fix the system is to prioritize websites that are there to share knowledge, not websites with their primary priority to make ad revenue.”

The best way to fix the system is to prioritize the best content irrespective of monetization strategy.

Search engines should not penalize websites trying to make money.

They should penalize spammy websites that add little value but know how to game the system.

The degree to which the creator monetizes their content should bear little weight with a search engine that is trying to serve the best content.

Re: Reddit can't build a better search engine

#217

> I was too young to use Google when it first got started, but according to many, in the beginning, Google had better or more accurate search results. I'm old enough to remember when Google came, and the main difference was this: With Google you did not have to flip through 3 pages of completely unrelated search results; that were paid for. Google used to display advertisements on the right as well, i.e. not among th…

The early search engines (Webcrawler, Lycos, ...) ranked "relevance" based on some silly metric like "the more times the search term appears on the page, the more relevant the page must be". So the result was that the first pages were full spam sites that just used search terms as hidden text. Then Altavista came and it was a huge change for the better. They at least made some effort of trying to rank actual relevanc…

> made Google what it is today is PageRank

Yes, but it was no secret that Google's page ranking was based on counting links from different web-pages; I think Page/Brin even wrote a paper about this when Google was in its infancy. But Google clearly had better ranking at first, but it would not help the others if their ranking was better than Google's, since the "paid results" polluted their search results so much they became a joke. A lot of the paid results was "adult content" as well, creating a somewhat strange atmosphere during presentations if the lecturer had to preform an internet search. Although a large part of the www was adult content in the late 90's :-)

The competition doomed their own business by selling search results; they really did not play the long-game, but went for the quick cash. I guess most people would have...

Re: Reddit can't build a better search engine

#218
post #212

Earlier quoted context omitted.

Well there still seems to be a lot of deep content on the web, it's just that google doesn't show you any of it. You may, if you're lucky, get a Wikipedia box. And Wikipedia, for all its breadth, is incredibly shallow on most topics. I submit that the reason you aren't learning anything from google is that it just doesn't surface anything for you to learn from, but usually just the same dozen or so big sites, wikiped…

So why you think this particular website must be the top answer? I looked at googles results and and I disagree, they are not bad, e.g. actual program examples, videos and papers for example.

Besides the ads, the Wikipedia article, the next result I'm getting is this: https://www.geeksforgeeks.org/strongly-connected-components/

Which seems to mostly just be paraphrasing the Wikipedia article and the page I linked. The kent.edu-page seems aimed at teaching you the material, the geeskforgeeks-page seems primarily to focus on getting you through your introduction to computer science exam. It gives answers, not understanding.

Re: Reddit can't build a better search engine

#220
I built https://www.unscatter.com using Reddit to source links for the search index. Last year I added Twitter as another source.

I don't think it's a "better" search engine. It's a different lens through which to search. Reddit and Twitter are an information source for what people are talking about. This is why I limit my index to articles that have popped in my Reddit/Twitter input in the last 30 days, deleting anything older.

I've actually had it up for years now, just don't know what to do with it. Been focused on my career in IT rather than entrepreneurism because well, life. I can say just this morning I saw "Stanytsia Luhanska" pop as a trending term on the front page of Unscatter and at the time mainstream media has not picked up the story of the school being hit by Russian shells.

I think over all the quality results still come from Reddit. Twitter often gets gamed and I see content terms pop up in the trending list. However, Twitter overnight (my time, US East) gets a more international flavor with lots of Korean and other Asia Pac country content bubbling to the top during that time because of Twitter.

Post reply on HN