legolas

joined 10 months ago
[–] [email protected] 1 points 20 hours ago* (last edited 20 hours ago) (1 children)

I asked ChatGPT about existing experiments on this matter. Sheep behavious is real:

Key Studies: The "Hidden Votes" Experiment (Muchnik et al., 2013)

One of the most well-known studies on this topic was published in Science by Muchnik, Aral, and Taylor. Researchers manipulated upvotes and downvotes on a social news site (similar to Reddit) with 100,000+ users. When an initial upvote was artificially added to a post, it increased the likelihood that others would upvote it by 32%. Downvotes did not have the same effect—they were often corrected by other users. The experiment suggests strong social influence in voting behavior. "Bandwagon Effect in Online Voting" (Lorenz et al., 2011)

This study found that when people saw public votes before casting their own, they converged towards the majority opinion. The effect was particularly strong in subjective judgments, like ratings of art or music. YouTube and Social Proof (Salganik, Dodds, & Watts, 2006)

In a controlled music experiment, researchers manipulated download counts for different songs. Songs with high fake download numbers became even more popular, showing strong herding behavior. "Hidden Likes" Experiment on Instagram (2019)

Instagram conducted real-world A/B testing by removing visible like counts in several countries. Initial reports suggested reduced social pressure, but Meta has not released detailed statistics. What Happens When Votes Are Hidden? Some studies show that without visible votes, people rely more on personal judgment instead of following the herd. However, others found that herding still happens when other subtle signals (such as comments, engagement, or reposts) remain.

[–] [email protected] 1 points 2 days ago* (last edited 2 days ago)

Yeah, I edited the answer forgot about case when user wants to vote on many comments, check it out now and see if your questions still hold.

 

It was probably suggested before, but the topic is not raised too often so here goes the solution:

  1. We introduce two types of threads - normal and "controversial/heated/political" (probably there is a better name)

Normal type is for "technical" discussions, where the best answer is accepted as best by some very large percentage of people, lest say 90-99%. The treshold could be a matter of discussion, but you get the idea. So that would be questions like "How to fix dead radiator in PC", "Whats best way to do this or that"

Controversial is for discussion where there is potentially lot of disagreement, but also where there could be just some disagreement, but we want to hear other points of view. So all of the political things, questions about genders, etc, everythign that creates heated conversation. Probably could also be used for humourous topics.

The thread type is set while opening a thread, but it can be changed any time during the discussion by forum moderator

  1. We leave normal type discussion as they are on reddit/lemmy whatever. For controversial first when user enters the thread, all of the comments are sorted in random order. All of the comments vote scores are hidden. Now user casts votes in "one go". Only until they finish casting votes, other votes are visible for them. Changing already casted votes on these threads is NOT POSSIBLE. They unfortunately cant vote on new comments which were added after they voted. THey are only allowed to vote once. TBH users dont usually come back to threads after they visited them once so its not like we make some common behaviour impossible

This way we eliminate sheep behaviour and demand making their own decision by user. And we force user to be responsible for making a decision. Someone might argue that we sometimes change mind, but it doesnt matter, cause the number of times we change minds is really tiny and the gained changed behaviour is far more valuable. After everything is visible for user they can now sort by most popular comments which is now available.

That should be it. We also might introduce thread freezing if necessary.

What do you think? And also, since lemmy is open source, do you think there is a change that some bigger instance migght create a fork that introduces some of these changes as an experiment?

[–] [email protected] 2 points 1 week ago

Yeah I saw that chart, but this data comes from reddit itself, which tells me to be sceptic about it. They are public company now and they have to look good to investors.

[–] [email protected] 5 points 1 week ago

yup its what I was thinking as well. Reddit might deliberately run ai bots to fake traffic.

[–] [email protected] -1 points 1 week ago* (last edited 1 week ago) (1 children)

Damn you sound like bot you know that? No typos, perfect answer. Seriously. Can you prove you're a human? xd

[–] [email protected] 1 points 1 week ago* (last edited 1 week ago) (1 children)

https://www.youtube.com/watch?v=RSr_vwZGF2k This is what I watched. I base my opinion on this. Im not saying this is true. It just sounded legit enough and I didnt have time to research more. I will gladly follow some links that lead me to content that destroys this guys arguments

[–] [email protected] 4 points 1 week ago

Did they? According to their repo its still WIP https://github.com/huggingface/open-r1

[–] [email protected] 4 points 1 week ago* (last edited 1 week ago) (2 children)

well if they really are and methodology can be replicated, we are surely about to see some crazy number of deepseek comptention, cause imagine how many us companies in ai and finance sector exist out there that are in posession of even larger number of chips than chinese clamied to have trained their model on.

Although the question rises - if the methodology is so novel why would these folks make it opensource? Why would they share results of years of their work to the public losing their edge over competition? I dont understand.

Can somebody who actually knows how to read machine learning codebase tell us something about deepseek after reading their code?

[–] [email protected] 3 points 1 week ago* (last edited 1 week ago) (3 children)

Its just my opnion based on few sources I saw on the web. Should I attach them as links to the comment? I guess I could. But thats extra time which Im not sure I want to spend. Imagine the discussion where both sides provide links and sources to everything they say. Would be great? I guess? But at the same time would be very diffcult on both sides and time consuming. Nobody doest that in todays internet. Nobody ever did that in causal conversations. Not just internet acutally, in both real life and internet. Providing evidence is generally for court talk.

You are right. We are all on our own in pursue of truth. And with rise of AI and fake reality things are going to be crazier and crazier each year. Pair that also with the fact that our brains have limited storage capacity for information and knowledge and it doesnt look bright for humans. I stay optimistic though despite that.

[–] [email protected] 2 points 1 week ago

Yup. Thats internet nowadays. Full of comments like this. Cant do muich about it

[–] [email protected] -3 points 1 week ago (10 children)
[–] [email protected] 6 points 1 week ago* (last edited 1 week ago) (3 children)

So are these techiques so novel and breaktrough? Will we now have a burst of deepseek like models everywhere? Cause that's what absolutely should happen if the whole storey is true. I would assume there are dozens or even hundreds of companies in USA that are in a posession of similar number but surely more chips that Chinese folks claimed to trained their model on, especially in finance sector and just AI reserach focused.

 

Interacting on instances which are not yet discoverd and connected to instance where we are registered is pain in the ass. Copying URL into search bar? That is really repelling. I'm sure it was discussed before but I can't find this discussion. How to make it easier? Do devs need to make changes to protocol or it's enough and it can be done within the apps themselves? How close are we to the solution?

 

My lemmy users per capita guess: 1. Russia 2. Finland. I can understand Russia cause developers are from there but why Finland?

view more: next ›