In other news, I will be really happy in like fifteen years when everyone has broken down and admitted that Unicode is the only sane way to go and all the tools use it by default. At the moment it’s built into most everything, but it 1. is generally broken and 2. is not the default.

(Clarification: I mean Unicode implementations are broken, not the standard itself.)

Google just posted a great graph of the uptake of Unicode on the web:

This is really encouraging, it’s great to see Unicode taking a plurality (if not yet a majority or totality) share. I think I had my two concerns reversed, Unicode needs to be set as the default for everything before all the varying implementations work out their myriad bugs.

But that’s progressing along well. Python 3000 and Ruby 1.9 will use Unicode as a default, and more content-producing tools are picking it up. There’s still lots of random standards that don’t support it (think of metadata stuffed into mp3s and images) or only half-support it in the “well, it’s 8-bit clean…” way that can leave you guessing at encoding.

I don’t have anything profound to say, I just like Unicode and wanted to make my prediction public.

Me: Hey, did you hear about [random technical development]?
Googler: Yeah, it’s real cool. Especially for [slightly offbeat application].
Me: That’s an interesting idea. Hmm, especially if you wanted to use it for [related research topic].
Googler: …
Me: I mean, that would work out great. It would have an [interesting performance characteristic], especially for large datasets.
Googler: …
Me: Hmm, do you think it might fix [persnickety sub-problem] if you also use [other technique or product]? I think it’s likely…
Googler: [deafening silence and refusal of eye contact]
Me: Hello?
Googler: …braaaaaains…

OK, the last line is whimsy, but the rest is the verbatim compilation of a dozen conversations. Google employees get distracted by mentally rereading their NDA and figuring out what they can say, can’t say, and can’t say that they can’t say. It just gets worse if you keep talking and extrapolating (that is to say, holding a conversation) because you’re giving them O(n² interrelations + m potential suggestions from them) more topics to analyze. I’ve never quite managed to crash them, but I imagine it’s entirely possible given the NP nature of conversations.

The fellow coder who inspired this post by pointing out the blank stares to me must remain anonymous due to their recent zombification.