I usually filter my searches with "score:>100" or something similar to screen out low-quality results. I imagine many do the same. However, I've noticed that older, lower-quality posts manage to make it through this filter on account of them having more time to accrue votes. I'd like to be able to filter by score per unit time; however, this is likely too niche a case to warrant a change of it alone. How about an extendible framework which allows filtering by basic algebraic operations on numerical metatags?
I imagine the searches are passed to an SQL for result delivery. A typical search would be "male/male order:score" which would be passed to something more efficient than
SELECT * FROM database.table WHERE tags LIKE '%male/male%' ORDER BY score LIMIT 60
All that would be required to implement this is to add a metatag "lambda" which creates a dummy variable as follows: "male/male lambda:rate=score/date order:rate" passes to
SELECT *, score / DATEDIFF(CURRENTDATE(), date) AS rate FROM database.table WHERE tags LIKE '%male/male%' ORDER BY rate LIMIT 60
Support would only be required for those metatags which have a convenient representation as a numeric type, and later implementations could extend this to other types. I imagine whoever is using the description regex functionality would love it.
Alternatively, y'all could let us have at it with raw SQL queries. That'd be epic; I can imagine that enabling some cool data science stuff.
Updated by KiraNoot