People are most valuable asset, not content - content is aim, but you need people to make the content in the first place.— SpeedyGonsales, old mantra of experienced wikipedian :)
- This same rule apply whichever is the size of project, large wikis have large number of users, so lot of things is going fine, but large wikis also do experience drop of users (both old and new users leaving), which leads to decrease of number of new articles (simple metrics) and slows down increase of both quality and total sum of content (more complicated metrics, but more relevant in the long run).
- I know that (and most probably every long time Wikipedian knows that) for a long time, but interesting Wikimania discussion motivated me to write this down.
- Although metrics should be put in place to recognize trends and to try to stop evil trends (if such are present), metrics which discover exceptionally good and likewise exceptionally bad trends could be used to ask the users to share what makes them so successful or opposite of that.
- Maybe most of bad practices were put in place by good intentions - trying to solve some existing problem, but without foresight which will be the end result.
- Changing of bad practices sometimes leads to conflict with users which invented bad practices, and do not realize that something is not working. Is there need to see ship sinking to know that something is bad, or are we smart enough so we can measure things, talk with people, and correct our practices as soon as we see there is a need for correction?
- Examples - simple one for start - welcome messages to people after their first or nth edit usually pleases new users. Welcome messages to people who did not do any edit on your wiki project, but just clicked on a interwiki link can annoy some people. Solution is simple - ask users to give welcomes only to users after their first edit.
- TLDR - More complex example - communication templates. When doing recent changes patrol some users could feel the need to automatize whatever could be automatized, and to swap writing of messages with just clicking on appropriate templates. Although that approach could be used to some extent (some small set of templates can be used in combination with human "touch"), every try to minimize human touch necessarily leads to almost exponential increase of number of communication templates, and then you need to be machine to remember all that templates or you need ESWB (extra smart wiki bot - software which does not exist) which will lead you fast enough using some algorithm like binary search to necessary template out of half a million of templates. So to be both efficient and kind with new users you will first spend time to make so vast a repository of templates that nobody will surely know how to navigate them even with software, and then you need to write the software to manage this mess of templates. Disclaimer - this is not so fictional example, but also - this is not accurate description of any wikipedia I know.