Having 150/160 points is too punitive

[xvrh]

When a package has 150/160 points, it can be out-ranked by other packages that have significantly fewer likes or downloads (200 times less).

Example: search for webview

Sometime it is impossible or very difficult to achieve the maximum score 160/160.
For example, the version of the Dart formatter used by pana can be slightly different from the one used within the package, preventing the final few points from being awarded.

I think the ranking should treat 150/160 the same as 160/160.

Also, note in the screenshot that the exact name match is a bit useless and could be removed.

[sigurdm]

For example, the version of the Dart formatter used by pana can be slightly different from the one used within the package, preventing the final few points from being awarded.

That would be a bug in the formatter.
It is supposed to always format packages according to their language version (decided by the sdk constraint) - such that even a newer formatter still would format old packages the same.

Also, note in the screenshot that the exact name match is a bit useless and could be removed.

If you are searching exactly for the "webview" package it is desirable to show that it exists, even if it is outdated.
We have iterated a lot on the mechanism here, and it is very hard to always please everybody.
If you have a very specific idea for improvement of this, then please open a new issue.

[sigurdm]

We kind'a agree that the ranking seems off here. We should see if we can tweak the weights or something such that such a high difference in download counts and likes would reverse the ranking here. But it is a fiddly process.

[sigurdm]

I don't think treating 150 and 160 points as exactly the same would be beneficial, then we should not really score at all.

But maybe taking the square root of the normalized score would be more reasonable, then the last few points would count less than the first.