Really beautiful product, but the way I see it there is a fundamental issue with the ranking (please correct me if I'm wrong): You are mixing up correlation and causation. It isn't ranking influential Product Hunters, but rather "Product Hunters with mainstream taste that tend to vote early in the day" (favoring European users that log on before lunch).
Am I missing something? If not, how will you deal with this issue?
@kjemperud thx for your valuable input! of course, European users have a better chance to receive a bigger score - but this is something we can't address on our own since we're more or less bound to the timezone ph is using.
we started the project one and a half week ago and tested 3 different approaches and found the current to be the best. what formula or criteria would you prefer?
@jollife I don't have the answer to how you should do it. It's a challenge, and might require some more advanced weighting of various variables, such as who else upvotes the same products, and their typical influence. This would not be easy to get right, but currently the scores are rather meaningless (other than for identifying mainstream European users). But I'm guessing this is a side project, so might not be worth dealing with.
Again, beautiful product. I especially like the comparison tool :)
@kjemperud yes, it started as a small side project, but our aim was a leaderboard, which everyone can really use. we will tweak the underlying formula over time, so please check back in couple of days!
hi @rrhoover. thx for your kind words! We've seen HunterData in the past and wanted to create an even better ranking.
How? Well, we incorporate two basic principles in the rating:
1. a user who votes early should always get a higher score as all subsequent voters
2. a user who votes on every single product should be slowed down by the algorithm. so we defined that votes for unpopular hunts decrease your rank.
another important feature for us is the ability to compare two ore more hunters against each other (see: https://uberhunters.com/all/rrho...). We loved to implement this feature, as @neuling2k and I tend to battle against each others all the time! :)
@rrhoover@jollife I'm not sure how much I like that votes on unpopular hunts hurt your ranking. Or maybe I use PH differently, but just because something is popular doesn't mean that it's good. I'm curious — if I went back and voted for all the top 3 most hunted items every day, would my score go up?
@rrhoover@jollife@dshan this is basically what @golocalapps is saying below. It's really hard to judge a product by name and tagline alone, and yet I'm guessing many people do. There are several different *types* of PH hunters that are useful: the original posters, the upvoters, and the commentators. There are some overlap between them, but it'd be interesting to see who are the most influential in those categories, rather than a universal metric based on who votes with the majority.
@rrhoover@chrismessina hi chris, that depends on the timing of your vote. if you vote for the top 3, when they all have little to no votes, than it would be great for your score. if you vote on the end of the day, where your vote is one of the last, than your score will not profit that much.
Very well done, @neuling2k and @jollife! HunterData, a leaderboard of product hunters, was one of the first (maybe THE first) community built product by @yvoschaap a year ago.
Truth is, the aggregate number of upvotes presented for each product isn't the best indication of influence or a "good hunt" because it doesn't account for the number of people that saw the product (our traffic has significantly grown over time and time of day can make a difference) or non-credible upvotes (from voting rings, for example). That said, this is a good first step.
P.S. I shared some data from 2014, including some of the "top hunters" based on a weighted average of the number of upvotes they received from posts made.
Teleport
Swat.io
Teleport
Swat.io
Swat.io
Raycast
Product Hunt Regular
Raycast
Swat.io
Raycast
Product Hunt