- From: Dominique Hazael-Massieux <dom@w3.org>
- Date: Wed, 14 Nov 2018 12:39:30 +0100
- To: Stefan Schumacher <stefan@duckflight.de>, w3c-translators@w3.org
Le 14/11/2018 à 12:06, Stefan Schumacher a écrit : > All I got to read was, that there were computerized translations that > were announced as proper translations and they were just used as a > traffic generator. > > I haven't heard of a case that there was a "good" translation and it was > replaced by something else. Was there? I believe that has happened, indeed. > If a bad translation got replaced > the problem already existed before a "changing file problem". > > Maybe @Dontcallmedom can shine a light on what type the massive abuse > was about. I didn't really get it. And what was the ratio? good/bad, > 50/50, 20/80, 80/20 or 95/5? I'm afraid I don't have detailed statistics, but I would guess 80/20 is probably already overestimating the actual detected abuse. That being said, I think when we stopped the program, the ratio was probably going up since it presumably had been identified as a working scheme. > So to get back to your question. > > You can validate a static file, but how do you do that for an HTML-Page > on the web. > Is that static file actually shown like expected in your browser if > someone uses e.g. JS to alter the content? An automatic process might > not help, and then it is back to manual work. > > So from my point of view that is all too complicated, all that technical > programatical approach. I feel, W3C and the translators should find some > old fashioned way to establish some trust. A personal conversation and > knowing who the other is, is not that difficult to accomplish. > (For that some invitation to some (yearly) event that took place anyway > was a nice idea in the past.) > A well maintained website of a translator is also a hint, if there is > any abuse to be expected. > > So instead of making it all technical, I ask for a small validation > list. Meet like two or three of, let's say, six criteria, and you are > good to go. > > Criteria might be: > 1. personal identification > 2. personal conversation > 3. well maintained website related to W3C cotent, technical content, > spec related > 4. someone trusted individual referred the person > 5. a W3C member is taking responsibility to watch over the translation > process and maintainance > 6. ...please fill in. Once concern with this approach is that it assumes trust established once remains true over time. It may also set the bar too high for some genuine good-faith contributors. Compared to the proposed new approach, it also doesn't really help with maintenance of translations over time, nor with their persistence. I'm quite clearly hearing pushback or at least concerns on my proposed approach, though; it would be helpful to characterize a bit more the nature of the pushback: * not wanting to gate translations through a review system * not wanting to get one's translations published directly by W3C (vs on one's own web site) * not wanting to use github for managing translations There seems to be a mix of "overengineering which takes away the fun of translation", of "removing some of the incentives that ones gains from exposure" - understanding this a bit more would help me assess which alternatives to explore or test for as we set up our new infrastructure. Thank you all for your input! Dom
Received on Wednesday, 14 November 2018 11:39:33 UTC