Stop prioritizing high-volume-output websites/accounts; rate content on merit instead

One purpose of the sorv algorithm is to rate each individual piece of content based on merit, without regard to how frequently the same author publishes other pieces of content. In other words, if blog X publishes a new piece of content every day, and blog Y publishes a new article once a month, and if blog X and blog Y publish an article on the same topic, and a random sample of the target audience rates blog Y's article higher than blog X's article, then blog Y's article should be given priority in search results.

But this raises the question: If someone produces a lot of content -- especially if they give it away for free -- shouldn't they be rewarded just on that basis, even if each individual article might not be as good as some other freely available article on the same topic? The answer, under most reasonable assumptions, is No.

This is counter-intuitive if you compare it to other scenarios -- for example, suppose two chefs volunteer to make feels to feed a neighborhood, and the first chef makes 100 meals that the average person would rate a 7, and the second chef makes 3 meals that the average person would rate a 9. By any reasonable metric, the first chef has done more good.

Here is the key difference: A chef who makes a meal (even if it's a 7) is giving a meal to someone who otherwise wouldn't have gotten the meal (assuming the total number of donated meals does not exceed the number of hungry people in the neighborhood). The recipient would rather have the 9, but the 7 is better than nothing. On the other hand, the person who writes a tutorial that rates a 7, is potentially displacing another tutorial from the top of the search engine rankings, if the search engine ranks the 7-tutorial higher than a 9-tutorial -- because the 9-tutorial can potentially be distributed to an infinite number of people, but the 7-tutorial is getting in the way.

Under the following idealized assumptions:

an author produces 100 tutorials, each of which would be rated a 7 by the target audience;
for each topic covered by that author, there is at least one other tutorial that the target audience would rate at a 8 or 9;
a user looking for a tutorial will read only one tutorial on the topic: the one that is returned at the top of the search engine results

then, despite the best of intentions, the high-volume writer's impact on those users has been negative.

Now, those assumptions are each unrealistically simplified, but if you replace them with more realistic assumptions, they do not affect the core argument:

People's opinions vary, and just because the average user rates a tutorial a 7 does not mean that all users will. Yes, but -- going by previous ratings is the best that you can do. If one tutorial has a current average rating of 7 and the other has a current average rating of 9, it would be ridiculous to recommend to a user the one that has an average of 7.
Users will likely seek out more than one tutorial. Maybe they didn't understand the first one; maybe they remember information better if it's repeated in different ways from multiple sources. Yes, but -- even if the average user might read three or four tutorials to learn a topic, those three or four should still be the highest-ranked possible, based on random-sample-voting. There is no reason why a tutorial should be recommended above others just because the author produces a lot of output.
Users may not have enough information to determine if a tutorial is "good". For example, if the tutorial is a recipe, users would have to actually follow the recipe steps to determine if it works; some users might not bother, and would simply give it a "high rating" because the pictures look pretty. This is a valid criticism, but it's true whether or not the author has a high-volume output history. (Either the user's ratings are a valid guide to quality, or they're not. If they are, then if a high-volume author produces a new recipe that gets a 7, and a newbie author produces a recipe that gets a 9, why shouldn't search engines favor the newbie's recipe? If ratings are not valid, then why should it matter if the author if the already produced 20 recipes with high average ratings? Maybe those raters were just handing out high ratings because they liked the pictures too, and the recipes don't actually work!)

Either way, the solution is not to favor high-volume authors but for the system to support "high-cost voting" -- where the initial random sample of voters is expected to do more work than is normally required for voting (e.g., to actually try the recipe steps), and they might have to be compensated (with extra visibility on the site, or with "points" awarded by the site, or if all else fails, with money).

On the other hand, a high-volume author might object that their content should be favored because as a result of producing lots of output, they have gotten objectively better. But if the content they are producing is good-quality anyway, this will be reflected in the ratings that it gets as a result of the random-sample-voting. There is no reason to give the high-volume author an advantage beyond that.

The argument above gives the reasons why taking away the advantages of high-volume content creators, and having all pieces of content compete on an even playing field, would result in higher quality content for users. There are also reasons why this would be a better experience for content creators:

A content creator would be able to find out, from the very beginning, whether the effort to create content produced enough benefit (in terms of recognition, "Internet points", or even money) to be worth it. On an uneven playing field that favors high-volume creators, a content creator would have to spend weeks or months producing a high enough volume of output to be noticed by the algorithm, in order to find out if the ongoing content production produced enough benefit.
A content creator would get accurate feedback on whether they were improving. In the current system which primarily rewards (a) high-volume output and (b) luck, a person who creates lots of content might be getting objectively better in the process, but unless they get lucky with the algorithm, they might never get enough traction to realize it. (Conversely, their content might not ever increase in quality, but if the algorithm boosts them because they produce lots of output, they might mistakenly believe it's because they're getting better!) With random-sample-voting, with no bias towards high-volume content creators, each piece of content would be rewarded in proportion to its quality, so the creator could tell whether they were improving (or whether they get their best results only from certain types of content, etc.).
A content creator would be less dependent on their "following", which means they would be less vulnerable to being swindled by the platform. For example, if a platform punishes a creator for violating the platform's "rules" even though the platform made a mistake, if a creator has built up a following on that platform, they might have little choice but to comply anyway, since it would be too hard to start over on another platform. (Similarly, if a high-volume creator earns money, the platform then just starts lowering the creator's payout for no reason except "they can", because it would take too long for the creator to start from scratch somewhere else.)

There's really only one category of user who would be worse off if sites started rating content by merit (independent of high-volume output), and that's the high-volume content creators whose content would be rated lower than other people's content in a merit-based system.

And yet, even some of them might find it interesting to try. A creator like Taylor Swift or Mr. Beast has arguably conquered every contest in their field, except one: In a double-blind side-by-side comparison, how would their product do against someone else's? (The most successful creators in their field are often already so rich that there would be no risk of financial harm even if they "lost" to someone else.)

What if Taylor Swift arranged a contest where anyone in the world could write a song, and Swift could submit her own entries as well, and users could rate performances of the song using a random-sample-voting system? (The songs would have to be performed by third-party performers, to stop people from voting for Swift's song just because they recognize her singing it.) Surely her songs would rank very well, but would some complete unknowns end up submitting songs that would rank better? I think it's quite likely there are people at the top of their field who would be interested to find out how they would do in a double-blind test against up-and-comers.