Ideals

I thought it would be interesting to talk a little bit about the standards process given a few events that lined up last week in my mind...

Last week my colleague Rakhi wrote a great piece about how standardization at the edges sometimes (maybe even often) works backwards because there are fallable humans all along the way - writing the spec, interpreting the spec, writing tests, and so on. The very next day, another such case was brought to the CSS Working Group. A lot of people found this very interesting.

Around the same time, we published a new episode of Igalia Chats about Global Privacy Control, which, at the end of the day, is mainly a boolean header that a browser can send. As we discussed in the Chat, it probably feels suprising to learn that the efforts to do something like this wind up spanning _many years_ (there are some helpful links there to the history of Do Not Track as well) and what makes people believe that this time it will be different (that the pressure kind of came from outside first is a big one). Anyway, it's kind of fascinating if you don't know any of this stuff.

Then, on Thursday, the SpellCheckCustomDictionary proposal advanced to stage 2 in WHATWG. If you're not familliar with it, I wrote a short post about it not too long ago. The reason I mention this is that it's been stuck for months at stage 1, because we needed to talk a lot about Internationalization concerns.

This is a potentially important discussion, to be sure. However, only "potentially" so because while spell check has existed in browsers for more than 10 years, for the most part, the `lang` attribute remains unconsulted. Where it is, it's not really specified how that should work in terms of general spell checking.

To me, that feels worth talking about, because it sometimes feels like there are impractical gaps that it's not clear how to fill. On the one hand, it feels unreasonable to expect a small improvement to spell checking to be held to a higher standard than the mature specification and existing implementations. On the other hand, I fully agree that these are important questions. The difficulty, at some level, is that all of these uniquely important questions are also somehow dramatically underfunded. Accessibility, Internationalization, Privacy - or even just integrating and normalizing MathML and SVG into One Platform being just a few obvious examples.

It's hard to live up to ideals if we're not funding them well enough, or if they aren't pragmatic enough (for example, demanding the the spec is right).

I don't really know how to fix the funding bit except to continue to lobby vendors to allocate more budget to those things (or, in come cases, maybe just any budget at all) and to suggest that we need more experiments in collectively funding and prioritizing those ideals in implementation work. Lots of features are like fast food or desserts - someone needs to make sure that we eat our vegetables.