Show newer
Andrew boosted

My wife says that any time someone proposes doing anything with an ML model, you should replace “AI” in the proposal with “trained weasels” and if it still sounds like a good idea you can go ahead with it.

Unexpected development: we did not find any significant amount of wet wipes!

We found a sock.

Someone flushed a sock down the toilet.

I ... have questions.

Show thread

Also, lesson learned: leave the lid down when going away. The result of an overflow is unpleasant but nothing like as bad as it would be otherwise...

Show thread

Just got back from two lovely days away to find my toilet is backed up, and upstairs's shower has thus unwittingly spread raw sewage over our bathroom floor.

The plumbing here is very weirdly wired up, but ultimately, this is going to be down to the horrific thing that is "flushable" wet wipes. As it was the two times before.

They're not. Don't.

Today's surplus-of-old-mysteries thought: it's noticeable how many of these:

a) would be solved trivially with tools of the modern era (phone logs, CCTV, DNA); but also

b) *are* solved by investigative methods that just seem ludicrously implausible now (eg routinely recording serials of banknotes)

Makes you wonder what the verdict in 2070 on contemporary mystery novels will be. "Strange to think of an era before we could check car route-finding logs"

It's a great Act from a special-cases perspective, to be honest. Not only does s.17 ensure that Shetland is not disrespected (though apparently you can box up Orkney or the Western Isles all you want), s.1(1)(a) has an urgent caveat to make sure they didn't accidentally omit Skye.

Show thread

Today's quiz question: a) what is the most unusually specific legislative requirement in the country, and b) why is it section 17 of the Islands (Scotland) Act 2018

Andrew boosted

Most folks I know who are worried about AI aren’t worried about these programs becoming self aware and oppressing us. That’s not a thing. We’re worried about all the ways these systems will ramp up existing inequalities and further obliterate our shared reality as human beings. That’s already happening.

Looking at a couple of randomly selected other groups: auto-assessed China-related stubs are 31% no longer stub tagged. For basketball, 42% (!)

A lot of the talkpage rating templates don't seem to track auto-assessed, so it's hard to come up with overall figures. But it does suggest our stub count is way overrated.

(I have suspected this for a long time...)

Show thread

Idle curiosity of the day: what does our quality-tagging lag look like?

There are 2819 "science and academia" biographies marked on the talkpage as auto-assessed stubs, indicating they were tagged as stubs at some point.

Of those, 1013 are no longer tagged as stubs: the main tag has been removed but the talkpage was not updated.

So 36% of "stubs" in this group were assessed at some point by a human as no longer being one. Hmm...

Chapter ten and he has not explained how to make the money yet, this is eerily up-to-date. all it needs is to add "27/?" and some random emoji

Show thread

I think this bit went around a couple of years ago from someone talking earnestly about tomatoes, now I think about it

Show thread

Today's serendipitous musical discovery: Bonnie Raitt covering 'Standing in the Doorway', absolutely lovely

at the moment I'm trying to get the Parliament linked-data IDs in place - currently on about 5000 with an ID out of 13500 identified Acts, versus 17500 in the Parliament source data. Another 7500 matches (same short title + year + chapter number) just going in, so by the end of the evening we'll have an attached ID for pretty much all the extant ones.

Show thread

This evening's work has drifted from into legislation: here is a report for all years since 1801 showing the number of distinct public Acts plus a rough indicator of where we may have gaps.

Broadly, decent coverage 1801-1870 but with a few gappy years; partial 1870-1960; pretty good after that.

Andrew boosted

I wish more people understood that "I want the computer to generate a natural language text that sounds like a plausible answer to a question about x" and "I want the computer to answer a question about x" are two very different problems.

Turns out "New Anzac-on-Sea" was a resort town on the Sussex coast set up in 1916 by a developer who bought a few acres of fields, and ran a contest to select the name for his new town: the contest itself seemed to more or less be a way to sell plots of land in it. After an outcry (and prodded along by the Act) it got renamed "Peacehaven".

Show thread

Update on this: the reason for the Act was even weirder than I assumed.

"It is not right or fitting ... that we should have "Anzac soap," and an "Anzac Motor Company," ... Worst of all, everybody will recollect that the word "Anzac" was used in the "Anzac-on-Sea" case where it was certainly put to a very base and improper use."

Show thread

This one doesn't actually tell us anything new (we've known about the Edgcumbe chain for a while and it's probably the longest there is), but pleased to find a different way of doing it without having to enumerate each generation in the code.

Show thread
Show older

The social network of the future: No ads, no corporate surveillance, ethical design, and decentralization! Own your data with Mastodon!