How would you calculate the quality of bait on 4chan? I was thinking the formula should go something along the lines of: Number of (you)s * The ratio of people thinking it's genuine. If you have any other way to calculate the quality of bait, may you please share it?
>>16532923A lot of (you)s is actually an indication of poor bait. The goal of bait is engagement, not dismissal.
the best bait are the posts that fall off page 10 with 0 replies or get pruned by jannies, because this means it caused such great seething that broke the mind of anyone who read it
>>16532923Sci and biz was way better in 2019-2020 the quality of posts is abysmal. Now it’s just dysgenics and midwits spewing bullshit redundant topics
>>16532923Do I get to see the IP logs, or do I need to work solely on publically available data?Am I to assume that I am OP in all the threads?
>>16534549/sci/ is now a training ground for science denialism groups beta testing memes.
>>16532923>How would you calculate the quality of bait on 4chan?Easy: it's 100%.
>>16532923It's the shape of the reply graph. Engagement per >>16532965 probably means long branching chains where "NTA" or styles create a conversation that survives the passing of the original authors. In other words great bait creates a transcendental circlejerk like Plato.
>>16534612This isn't a religion board.
>>16534805I want to yell at you for explaining my reply in detail, but you don't deserve it.(you gave it more thought than I did)
>>16534610Sorry for not specifying, this is assuming you join a thread and make a post in it. As well, this is assuming you are an average user so you aren't able to see ips of other users or the total amount. >>16532965>>16534805So you are telling me my definition of bait is flawed? And the formula I proposed is more akin to how much seethe you generate through a post. How would a new formula work if we take account the average length of these reply chains? The average length of a chain * the amount of (you)s? Or would it be more complex, I'm a first year in an engineering college so I don't really know how you would tackle this.
>>16535004>gets feedback>doesn't build upon the feedback>asks for spoonGet your own fucking spoon. The fuck are you getting an engineering degree for if you're going to pull this shit? Imagine scamming yourself out of a practical exercise(after having sufficient hints to figure it the fuck out handed to you)
>>16535021Alright, what if there was a value known as QoBt (quality of bait total) that is the total value of every individual QoB value per post. [QoBt= QoB1 + QoB2 + ... + QoBn]The formula formula for the individual QoB post would be something along the lines of: (You)s with replies divided by the total amount of (you)s [(Y)r/(Y)t] times the ratio of people who think it's genuine. [G/(Y)t]You could probably scrap the genuine part of the equation as well, and just get how engaging a single post is. This is as well combining all of the branches into its own node instead of tackling them separately. Hope this is in the right direction.
>>16535053I messed up the picture, pretty sure the image should look more like this. On the right is how the formula is supposed to look like.
>>16535058This also flew over my mind, but thinking about those /v/ posts with 30 replies, I've realized they would be considered really poor bait. So I added in the total amount of (you)s as a standalone element in the formula that will be manipulated by the percentage of (you)s that got a reply on their own and the percentage of people who think it's genuine.
>>16535004I still can't tell what the formula is. What's the most baity thread?
>>16534809>"Science" is completely faith based and you are no different than chanting monks.You forgot the image for your meme.
>>16532923Build the directed graph for the thread. That is, link y->x exists if post y replies to post x. This graph will be highly integrated. Compute the radiality of all nodes. Compare the radiality of the bait node to the radiality of all others to see if it was "more successful".Divide node radialities by overall graph radiality to create a percentage score to compare across graphs.
>>16535650The entire formula is, how many of the replies to your own post have their own replies. And then counting the posts that have a reply to the replies. Etc until you don't have anymore posts.And then you add up all of the numbers of the set. I though the G% could be used if there were a lot of fish and hook images to discount those posts from the pool. But >>16535905 has a better way to score this.
>>16532923Your equation just yealds the total number of people that believed the bait.Did I just get baited?
>>16532923Flair per updoot
>>16532923Just see how many upvotes it gets on /r/greentext. The better the bait, the more predditors like it.
>>16532923You have no way of knowing who thinks it is genuine. People play along for various reasons.>>16535053Clever but how are you reading minds?
>>16536229No, I'm just a midwit. I thought about it more, and like this is the final formula. Btw there's no sincerity factor because I can't read minds like how >>16536278 pointed out.But basically. I just came to the conclusion that the formula should be something like, the generations of a post * total replies * ratio of bait. But at this point, I think I'll just stick to stasis and metallurgy.And now that I think about it even more, I just restated the formula you said but I added how long the people thought it was bait for.
>>16532923KYS kiwitranny
>>16532923real bait has never been triedhttps://m.youtube.com/watch?v=cUR4up6EBLA
>>16532923I liked the classical format where a single grammatical error or misspelling was required, and when rating baits I would remove a single point out of ten for its omission.
>>16535078A bait just flew over my mind.
>>16536638Youch