Here's a thought: show the rating as a 90 day rolling average with the option to view the all-time stats. Throw in additional logic to handle the cases where there are too few ratings in the last 90 days... then use the last n ratings (for some reasonable value of n), no matter how far back they go, for example.

That would seem to me to give the best of both worlds. If you release a crappy app and then later release a glorious update, you'll have to earn back your new ratings over time as old ones fall off the chart which encourages developers to release quality apps starting from day one.

Yeah, it seems like a simple reset isn't valuable.. makes more sense to weight the ratings over history.. so that ratings on the new version matter (much) more than ratings on the old versions.. but old ratings still count at a lower weight.. that way in your case you'd still be fine releasing a v2, but for apps that are consistently awful their rating wouldn't be so easy to pull up.

Another idea: Since Apple knows exactly how much money I've spent in their store, that signal can somehow also be weighed in with my reviews.

This means that reviews by people that have spent a fair amount in the app store would get a better/heavier scores than those that have not yet spent as much money in the store.

This would be susceptible to the same problems as Chris mentions in his post:

> Companies like TapJoy let you pay to get in the Top 25, and then once you are there you can get “organic” downloads by being on the toplists.

If they can arrange to actually buy enough copies of an application to get into the Top 25, reviews posted by those accounts would surely weigh very highly under your proposed scheme as well.