It’s also torches and everything after the regulator, which run at much lower pressure. At least in NZ

I think it might be because they’re connected and disconnected regularly so misconnection is a common problem, even with colour coding. Gas work on houses involves actually putting the fittings on pipe and is done by people who should be concentrating more on that rather than on what they’re about to weld/cut.

I’ve heard flammable gas uses reverse (left hand) thread to prevent cross connection. At least for welding gases in NZ; not sure about natural gas.

Yeah, I have no idea either, but it’s been around for more than a decade so it should be fairly easy to find a library that duplicates it.

I would be wary of AI-based solutions. There’s a risk of it picking up e.g. satirical/spoof sponsorships as actual ads, and perhaps not detecting unusual ads.

I’m slightly terrified of the day someone starts getting AI to reword and read out individual ads for each stream.

You definitely would have legal issues redistributing the ad-free version.

Sponsor block works partly because it simply automates something the user is already allowed to do - it’s legally very safe. No modification or distribution of the source file is necessary, only some metadata.

It’s an approach that works against the one-off sponsorships read by the actual performers, but isn’t effective against ads dynamically inserted by the download server.

One option could be to crowdsource a database of signatures of audio ads, Shazam style. This could then be used by software controlled by the user (c.f. SB browser extension) to detect the ads and skip them, or have the software cut the ads out of files the user had legitimately downloaded, regardless of which podcast or where the ads appear.

Sponsorships by the actual content producers could then be handled in the same way as SB: check the podcast ID and total track length is right (to ensure no ads were missed) then flag and skip certain timestamps.