For why I think subscribers play an important role, even in the Popular ranking:
If you go through Workshop items, ordered by the current Popular ranking, you will see wildly different download counts, and often even broken mods.
The only very consistent trend you notice is that subscriber number steadily decreases, without major exceptions.

This heavily suggests the current ranking strongly factors in subscriber numbers, which is a problem because the Workshop stopped keeping track of these a long time ago.
Because of this, users get very outdated and often irrelevant suggestions, featuring even broken mods in the very top. At the same time, it's very hard to discover new mods, since they have a major disadvantage in the ranking.

Any change to this would already be a huge improvements, but just to provide an example, here's one thing I would suggest:

• Change the Popular ranking to be mostly or entirely based on the unique download count within the last 7 days
• Replace the Subscriber ranking with an all-time popular ranking, which is based on the total unique download count

Unique downloads here are fresh downloads by users, so excluding update downloads. This makes sure mods are actually ranked by their popularity, and not by how often the author pushes updates.
While this data might not be available in retrospect, you could make the first update after this change, or the presence of the mod, count as a fresh download, and achieve the same effect this way. But any change, even if still based on any kind of download, would already much better than the current system.