Not sure if the article above is the best explanation of Steemit... but going to take this opportunity to contrast it with Relevant's cryptoeconomic design.
To start, Steemit has definitely pioneered the space and is the first 'curation market' not to mention one of the biggest dapps out there. Relevant economic design was inspired by that of Steemit. Here is how its different:
The underlying assumption of curation markets like Steemit or Relevant is that you can use market dynamics to solve some of the content curation and distribution problems we see on the web today. But in order for the market to actually be useful it needs to have some strong fundamentals, otherwise all you'll get is speculative noise. A prediction market is a good example. The reason prediction markets work is because the resolution of the market (an event that tells you whether you won or lost your bet) is a real-world event everyone can observe. This is a strong fundamental.
The market design of Steemit doesn't have this. On Steemit you bet on a post's success by staking tokens on it, but success itself is determined by how many tokens are staked on a post. This is a feedback loop. The bet or prediction is no longer a bet but a self-fulfilling prophecy. The market, as a result, is not based on fundamentals but on speculation. Any useful information the market might provide is drowned out by the noise of speculative activity.
This is why designed the Relevant economic model to be a proper market. On Relevant there is a distinction between a prediction (or bet) and the outcome. Like on Steemit, with each upvote, you stake a portion of your tokens as a bet on the post's ranking. However the ranking is not determined by the amount of tokens staked, but by the reputation of the voters (via a sybil-resistant pagerank algorithm, but that's a different topic). The two mechanisms are independent. This allows for a proper non-speculative market to emerge. Predictions now carry a useful signal because it doesn't make sense to arbitrarily pump a post to try to win some rewards — only posts that get vetted by reputable users are able receive payouts.
This enables us to construct a filtering funnel UN_FILTERED (New) -> FILTERED_BY_STAKE (Recommendations*) -> FILTERED_BY_REPUTATION (Top). Users with no reputation (or even recommendation bots) can stake on posts as an initial filter and suggestion to reputable curators to examine. Once a posts gets upvoted by a few reputable users it graduates to the final stage and gets a ranking in the 'Top' feed.
*You might have noticed we don't have a 'Recommendations' feed yet — we're working on it!