Ranking algorithm based on Upvotes and Views, without downvotes [closed]

Question

I'm building an app where videos can be 'Liked' (upvoted), and we're tracking unique views, but there is no downvoting.

This article seems to outline the standard for ranking videos that can be both upvoted and downvoted, preventing early submissions from dominating based on their seniority. However, the math is a bit too advanced for me to understand whether or not equating views (which might count as 'apathy votes') with downvotes for this purpose will still breed useful results. This will still reflect the ratio of upvotes, but may well punish a video for being popular in terms of views without garnering a bunch of likes too.

Do you expect that I could use this algorithm, equating views with downvotes?

If not, how might I go about ranking based on the ratio and quantity of views and upvotes only?

Thanks.

Answer 1

Using views for down votes would pretty much break his algorithm. Hes using a confidence interval(95%) of a Bernoulli trail. In Bernoulli trails your outcome can only be one of two possibilities(in this case either a 'upvote' or a 'downvote'). It may still work in the sense that you could redefine the outcome as either 'they came to the page and voted' or 'they came and didn't vote', in which case your p would be the number of votes/views and the n is the number of total views. I'm not sure how well this system will work for ratings though, as a item with 1 view and no votes will rank higher than a site with 1000 views and no votes.

To me, it doesn't seem like a good idea to penalize for views. Sometimes simpler is better. I would probably rank on votes per week or something like that.