I have recently been accused by Danny of using machine translation to create new articles on the Tibetan Wikipedia.

While I have been trying to cut back on the number of e-mails I send to wikipedia-l nowadays, I think this is fairly important.

I would like to make a few points:

1. TTBOMK, there currently exists no machine translator which will translate to or from Tibetan.

2. I personally recognise that current machine translation technology is generally unreliable, although it is improving a little, and would under no circumstances write an article using the unedited output of a machine translator, for any language. Nor would I use a word-by-word translation from a dictionary as that is even more unreliable.

3. All the content there contributed with my username is written by me. I am not fluent in Tibetan, but I know a little. I did not go beyond what I am comfortable with writing, and I believe it is all correct and if it is not, that there is much less error than if it were a machine translation. (the exception being my userpage, where I just wrote about myself, which I'm pretty sure is a bit wacky but hopefully not completely incomprehensible). The individual words in the table of contents on the Main Page, I used a dictionary for (after all, they *are* individual words).

Even if I made some mistakes, that would be even more encouraging to a fluent speaker to edit it and fix it than if it had no mistakes, and that might lead to more editing.

In addition, apparently the problem with "spam" on Wikipedias is, at least according to Danny, generally quickly cleaned up every time it occurs, with the exception of squatting.

If squatting is the problem here, locking Wikis surely is not the solution:

If I wanted to squat on a Wiki, I could almost as easily "show interest" and ask for it to be unlocked and start my squatting as if it weren't locked at all - the squatters on the Nauruan Wikipedia were committed enough to squatting that they asked a developer to change the name of the language on interwiki links to "Nauroese", and (I'm not sure about this second part, I didn't check) request adminship. Requiring a minimum number of committed contributors won't help either, as if somebody wants to use a Wikipedia for a conlang not likely to be approved for a new Wikipedia, there is a good chance they can find 4 other people to support them and contribute (especially if the conlang has fans or supporters, as do for example the Rosenfelder conlangs).

There are three solutions to the (perceived; so far it has only for sure occured once on any Wikipedia, perhaps twice) problem of squatting:

1. Verify in some way that the content people write is in the language they claim it's in, which does not nessecitate locking.

2. Erase all inactive Wikipedias, and never create any new ones.

3. Continue the way things were before, where people patrolled such inactive Wikipedias. This has until now been 100% effective against squatting - I caught the only known case, and another suspected case (the suspected case being relatively soon after it was edited). If there are other cases already, I have not caught them, so I guess the 100% statistic is invalid as if a case is not caught, it cannot be counted as "not caught".

In addition, Danny has suggested that I do not in fact check inactive Wikipedias on a regular basis. I, as well as a few other users, do actually check inactive Wikipedias on a regular basis. Some have expressed the feeling that if such Wikis could simply be locked instead, then it would save them the trouble of having to monitor them. However, I (and undoubtedly at least a couple of others) do it for different reasons than these people, and actually enjoy performing that task.

Another solution to a supposed problem of inactive Wikis is to actively recruit people to contribute. So far, I have done this on a very small scale with a small degree of success, and I have commitments from people to contribute to a Wikipedia once they finish something they're in the middle of. Presumably, if it were on a more official level and from an e-mail address not from a free email provider (I've begun to suspect that some ISPs block gmail addresses, after I've found that I get one reply per 10 or so languages, even though the number of failure notices I get is extremely low), the response would be greater.

In addition, I have personally found so far that there are at least a few Wikipedias where we already have Wikipedians who are fluent in the language but do not contribute to it because they either don't know of its existance or are busier with another Wikipedia. This is true with some of the Wikipedias in South African languages (people contribute instead to the English or Afrikaans Wikipedias), some of the Wikipedias in Indic languages (people mostly contribute to hi: and ur: instead of their local language with the major exceptions being ta:, kn:, and ks: and to a smaller degree a couple of others), European minority languages (I have a strong suspicion that there are at least a couple of speakers of French minority languages on the French Wikipedia who do not contribute to that Wikipedia, same with Russian and Italian minority languages, and with Saami speakers on Scandinavian-language Wikipedias). The problem of simple lack of awareness would be fairly easy to remedy, but if somebody doesn't want to work on a Wikipedia in their own language, there is no forcing them.

Mark