I'd not call per edit notifications infeasible, but without careful rollups they would be spammy.  The issue I have with them is they would not be very useful or motivating as raw notifications from a naive implementation except to very new or casual editors. 

Noticing that an article I improved got more popular or that a topic I've been interested enough in to contribute to has been mentioned in the media and suddenly spiked would be great notifications, but without a lot of developer work  hints at those would get drowned out by background noise of "Already popular/controversial article you corrected a typo in once has had another 1000 edits". 

Rate of change notifications would be exciting but very hard to provide.  Counters would reasonably easy to provide but noise to most editors.

I like the "people from 5 continents have read the article you created/edited".  It would come at a reasonable volume for most editor usage patterns, and could happily be processed offline once per day as it is purely intended as positive reinforcement. There is no response required so it need not have a realtime feel.

"People from 50 countries have read the article you created/edited" falls in the same bucket.

Luke


On Tue, Feb 5, 2013 at 2:14 PM, Ryan Kaldari <rkaldari@wikimedia.org> wrote:
On 2/5/13 10:01 AM, Fabrice Florin wrote:
Given these criteria, the most promising ideas so far appear to be notifications about contributions (or contributors) to a page you edited. We might also consider 'mark as useful' or 'feedback' notifications for projects that seem likely to adopt these new features, but those would have to be viewed as lower priority than the contribution notifications. Other ideas may have to wait until future releases, sadly.

Generating Echo events for every edit on Wikipedia for every editor of every article (even if they are bundled or filtered to a certain group of users) would quickly explode the database. It would also likely crash the servers with massive numbers of expensive queries – for every edit, we would have to look at all previous edits, and determine if those editors (which could number in the thousands) should receive a notification. Considering that we average more than 3 edits per second and have limited processing power (saving edits already takes 30 seconds or longer on long articles), I'm not sure this would be feasible. It also overlaps with existing watchlist functionality. Of all the ideas proposed, I think the contribution notifications are the least feasible and the least useful. The 'mark edit as helpful' idea seems like the best one to me so far, but would probably make more sense for E3 to develop than E2 (so that they could experiment with different UX and A/B test them).

Ryan Kaldari


_______________________________________________
EE mailing list
EE@lists.wikimedia.org
https://lists.wikimedia.org/mailman/listinfo/ee