top of page

Updates on Google Translate and the California Recall

In the title text of xkcd #1051, "Visited," Randall Munroe says "I hate when I read something like '... tension among the BASE jumpers nearly led to wingsuit combat ...', and I get excited because 'wingsuit combat' is underlined, only to find that it's just separate links to the 'wingsuit' and 'combat' articles." This post is about Google Translate, and about the California recall, but separately. If you're like Munroe, you might be underwhelmed. My apologies.

Just a short post today. Two quick but important updates to previous blog posts I've written this year: grapes; for the worms shall eat them and The Three Mysteries of the 2021 California Gubernatorial Recall Election. If you haven't read them, you probably should now. Anyway, on with the blog post!

The Part on Google Translate

Some bittersweet news: Google Translate fixed grapes; for the worms shall eat them.

I can explain. In my blog post "grapes; for the worms shall eat them," I said that when you Google Translate worms into Latin and back, you get grapes;, repeating the process gets you to grapes; for the worms shall eat them, and continuing leads down a bizarre trail of biblical-esque weirdness. Read the post, it's great.

But now, if you translate worms in and back, you just get worms.

Many of the other examples from the post also no longer work. grapes; in and back goes to grapes, then raisin, then raisins, but that kinda makes sense.

I think it's part of an across-the-board improvement to Google Translate Latin. Below is an example of four sentences showing how simple substitutions of names totally changed Translate's syntax (from Stack Exchange–I touched on this more in the original post)...and then what those sentences translate to now.

Not perfect, but much closer.

My feelings about all of this are mixed. This looks like a huge improvement to Google Translate, and will greatly aid anyone who needs it for a real-world task, if for some reason they need to get the gist of a Latin passage. But, of course, it was real fun to make fun of Translate while it lasted.

(Oh, by the way, before it was fixed, the grapes; for the worms shall eat them text did change a fair bit. It apparently contained such phrases as the glory of serpents, Silence: the root of all evils, and his sons shall Version: Mark: be full of worms, and ultimately stabilized on because it shall be wasted with worms; This says :: crop failure cancer and worms: the loneliness of age, some waiting, and they smell bad, and they were later released and no. I only heard about this afterward, so I wasn't able to try this, though.)

The Part on the California Recall

In The Three Mysteries of the 2021 California Gubernatorial Recall Election were, of course, three mysteries. The third was on why the recall was happening, and it had a pretty clear answer. The second, why no real Democrats were running, and I gave something of an answer as well. (Though this mystery was touched on more clearly in the Astral Codex Ten post "If You're So Smart, Why Aren't You Governor of California?")

And then there's the first mystery, the one I started and ended with: why is the recall election so close? At the time I wrote the post, the FiveThirtyEight polling average had the recall losing by only 1.2%.

Well, now the recall's down by 10.5%, according to that same polling average.

This makes much more sense in a state like California. It's still a little low, to be honest, given that most Democrats win statewide elections here by about 25%. But it makes me more willing to accept the unique situation of the recall election as an explanation for the difference.

It looks like I published my blog post at something of a high watermark for the recall effort. That weird SurveyUSA poll...looks far more like a fluke.

SurveyUSA even published another poll at the end of August, which had "keep" up by 10. That's a 19-point swing from their previous poll! Weird.

Republican Larry Elder is still winning the replacement race comfortably, by the way (presumably because Democrats are being told not to vote, which I mentioned in my previous blog post), though it's looking more like the victory will be irrelevant.

I still think some caution is in order. When polls are wrong, they're often wrong systemically. It's not implausible that Gavin Newsom could be recalled anyway. PredictIt now says that the recall has a 14% chance of success. I think that's reasonable.

As always, we'll see.


I'm not sure if there's a common thread between the fall of the grapes and the change in recall polling. My previous blog posts' theses being undermined? I'm not sure if it's as extreme as that.

Regardless, thanks for reading my two blog-posts-in-one about two very different, unrelated topics.



Related Posts

See All
bottom of page