Platonides wrote:
I find it a bad example, as it's not clear what you are translating from the keys (and there's no documentation on them) a) en/zh/ca... mean "The name of the article about lang XX on XX.wikipedia.org", not the language name. b) "Good" means "articles" c) "Administrator" means "sysop" d) 'Template:Statistics' is translated, but translations don't exist. And if you started creating them you'd face opposition by the community.
Other than the last two messages, IMHO they could be automatically generated.
I think that "Good" and "Administrator(s)" were in original table on en.wn. Also, there are two types of articles "good" and "total", so there should be written "good articles" from which "good" is an aberration.
However, I changed some of the messages exactly because of automatic translation via Wikipedia interwikis (instead of "Last update", I put "Time"). Except Polish (and English) all other translations are not checked by human. (I think that I changed "Template:Statistics" for Ukrainian, because it was obvious that the name of their article is "Statistics (science)" or something like that, like in Polish).
Also, I didn't make comments (or more descriptive names). However, it is assumed that a person who is correcting translation should know what is translating. And, yes, for some wider usage, there should be much better described. However, I was thinking about possibility to move bot localization to betawiki (but I still didn't talk with anyone about that). As I said, this bot is still in the stage "willing-to-be" multilingual bot :)
And, of course, I didn't create them without asking communities (sr: -- community more or less = me; I am (more or less) active on en.wn; and it may be said that I got permissions from pl community (asked on IRC)).
I find the interwiki links of the village pump extremely useful. Of course you may get it moved to their [[Wikipedia:Embassy]] or get a nice pointer to [[Wikipedia:Bots]]. So yes, it's time expensive but it think due to communicating with communities, waiting...
Last time (in April this year) when I wanted to ask a simple question all communities ("Do you write language names with starting capital letter?"), I found that it is extremely hard to find all village pumps, as well as *methods* for posting new section. The consequence was that I realized that I have to ask everyone for some basic data for their vps. Product is something like 2-3% covered village pumps [1]. The other page (but, without a "post method") is [2]. But, please, notice (1) the number of covered projects, (2) how many projects have a link to "discussion".
Also, the most of projects have explicit decision to make interwiki links from their village pump only to the village pumps of the most important (globally and/or regionally) projects.
There is another problem. Please, look at [1] what is the method of adding new post to vp at Russian Wikipedia. I remember that Danish Wikipedia had also unusual way for adding a comment...
So, theoretically, it should be a good idea, but practically it is still very problematic. Spending time on talk with a lot of communities is not a problem. Spending time on trying to talk with communities is a problem.
According to my experience, it is not physically possible only to ask all Wikipedias only one question during one full day, but I suppose that it is necessary to spend a couple of full days for that task.
The other issue, is, of course, fact that it is not reasonable to spend human time for something which may be done by computer. This includes one question * 700 (or, at least, * 250).
Some communities will block your block unless it has a bot flag. And i'd feel stupid asking a bot flag for a bot which posts "Hello World" ;)
I don't know for any bot which is blocked because of ~10 edits on its own page or subpages. Sense of blocking bots without bot flags is related to their high activity and contaminating RC, not to the fact that they are programs. (Of course, assuming that bot is operated by reasonable and constructive editor.)
Also, there are a lot of bots which are doing useful things. But, they are limited to one or, at most, a couple of projects. A lot of people are making bots which were already made a couple of times. And we have all prerequirements for making multilingual bots, except the basic multilingual methodology.
Agree. Maybe the need is not of a framework but a [[meta:Bot pump]] where you can request and provide bots.
There is a [[meta:Bot aid]] for such things. However, 10% more work on one bot (making a possibility for its localization and WM wide usage) means 700 times more useful bot. But, without those 10% more work, writing another bot for the same task may be the same job as previous. And, usually, the page was/is used for not so trivial asks and, usually, questions were not made by people from some of Wikimedian project.
So, my idea is to make a field for sharing methods, as well as for education of new bot programmers. Also, it should be used for promotion of automation, at least at a level of introducing non-tech people what bots are able to do.
Also, this shouldn't be the last project of this type. I treat it as "an academic introduction" to bots on Wikimedian projects (where all of participants are professors and students) :)
[1] - http://meta.wikimedia.org/wiki/List_of_Wikimedian_pubs [2] - http://meta.wikimedia.org/wiki/Meta:Index/Requests_and_proposals