MarketRank: The Anti-SEO Ranking Algorithm

And so, the oversimplified description of MarketRank is “just add up all the upvotes”. We have some more work to do to make the values accurate, but this is the general idea.

A ranking algorithm can be judged by the results it produces when people try to game it. What we need is a ranking algorithm that can only be gamed by creating genuinely good content. If “gaming the system” and “creating high quality content” are the same thing, then we have created a good algorithm.

It is unclear what proportion of the value of an upvote is quality vs. everything else. Finding a better way to extract quality from upvotes will be an important problem to work on moving forward.

The websites at the top of Google are not the highest quality websites, but the ones that put the most effort into SEO [1] [2]. There is an entire industry whose purpose is to game their way to the top of search results, which inevitably leads to those results being debased over time.

We need a ranking algorithm that cannot be easily gamed. This paper describes MarketRank, a ranking algorithm that is immune to SEO and simple to compute. We discuss the advantages and limitations of MarketRank, and show how it compares to Google’s rankings.

We claimed that upvotes were a good measure of quality but this is not necessarily true. In reality, an upvote on any platform is a combination of quality and other things such as the popularity of the author, quality of the discussion produced by the article, and many more.

No matter what Google’s search algorithm is doing to match the phrase “startup ideas”, we cannot make much sense of the fact that a page titled “How To Get Startup Ideas” on one of the most popular startup blogs, is considered by Google to be page 4 garbage.