Add a rating system to a library and you have not just added information, you have added a force. Ranked lists change what gets seen, what gets seen changes what gets voted on, and within a couple of years the top of the list is a self-portrait of the ranking mechanism rather than of the library. This is not a criticism of any particular site; it happens everywhere ratings are used for sorting.

What a rating actually measures

It measures the opinions of people who saw the item, which is not a random sample. Items that appeared on a front page early get thousands of votes; items that did not get twelve. The averages are then compared as if they were equivalent, which they are not in any statistical sense.

Early votes matter disproportionately because they determine visibility, and visibility determines later votes. A piece that got a good first fifty ratings can ride that position for years, while an equally good piece that launched on a quiet week never surfaces.

The effect compounds across sites. Because operators watch each other, an item ranked highly on one library often gets featured on another, which produces a second wave of votes from a fresh audience. What looks like independent confirmation is usually the same original accident propagating.

None of this requires anyone to game the system. It falls out of the arithmetic on its own.

Averages hide the interesting part

A four-star average can come from everyone thinking it is fine, or from half the readers loving it and half hating it. Those are completely different books and the average erases the difference. The distribution is where the useful signal is, and almost no site shows it.

Divisive work is exactly what a mood-led reader is often looking for. Something that half an audience found too slow is frequently the piece that rewards patience, which is the argument made on the mood-first end of the genre.

Where a vote count is visible, use it. Fifty votes at four stars tells you more than five thousand at 4.1, because the small number has not yet been flattened by the crowd.

Ranked lists narrow a community

If everyone enters through the same top-twenty, everyone reads the same twenty things and forms their taste from them. New readers then rate new work against that baseline, which reinforces it. Over years this produces a community with strong shared opinions and a shrinking sense of what else exists.

Newer readers feel this most. Arriving at a library with a decade of accumulated consensus means the first thirty things you see were chosen by people whose tastes formed under different conditions. There is nothing wrong with those choices; the problem is that they are presented as the library rather than as one slice of it.

It also shapes production. Creators watching a ranked list learn what the top of it looks like and make more of that. The feedback loop closes, and a whole category can quietly converge on one formula without anybody deciding to.

Sorting differently changes the library

SortWhat you get
Top ratedThe community's frozen consensus
Most viewedWhatever a front page featured
RecentUneven quality, genuine variety
By artistConsistent tone, best hit rate
RandomSurprisingly effective for breaking a rut

Recent and by-artist are the two worth defaulting to. Recent costs you a higher failure rate and buys you things nobody has told you to like. By-artist has the highest hit rate of any sort available, for reasons covered on how tag taxonomies really work.

Keeping your own scale

The most useful rating system is a private one. Note what you finished and what you abandoned, with one line about why. After twenty entries you will have a clearer model of your own taste than any site average can give you, and it does not drift with the crowd.

It also survives site closures, which in this field is not a small consideration. The habit is described for collections on building a gallery of your own.

Using a rated library without being trained by it

Read the vote count as well as the score, prefer distributions where you can find them, sort by recent or by artist, and keep your own notes. The ranked front page is then one input among several rather than the shape of your entire reading life.

And whatever the ranking says, the floor holds: fictional adults, unmistakably grown, nothing coercive, nothing taken from creators who said no. That is on the understanding of this house.