@simon
Yep, those are giant pumpkins all right. Absolute UNITS, every one!

@simon They look a bit Jabba the Hut, you know?

And this morning there are witches!

@simon I've found the self-reported Llama2 training (!) emissions interesting. See section 2.2.1 here:

https://arxiv.org/abs/2307.09288

That doesn't say much about inference, tho, but it's a start.

You'd also have to include not just GPU energy but e.g. emissions of the data center and supporting software around it, cluster utilization, the emissions of people working on this, traveling, and so on.

There's Scope1/2/3 etc. categories to make this a bit more structured but in the end it's tough.

@simon The part where you calculate your use‘s footprint makes sense (and would be very interesting to me, too) - justifying it by offsetting it against a specific unrelated „bad footprint“ activity like flying not so much unless you set a hard CO2 budget for yourself.

@simon Knowing the location of datacenters and using a source like https://app.electricitymaps.com/map you can guestimate the carbon footprint. My employer did the same for its datacenters in Norway but its somewhat complicated by a complex energy exchange market (we're connected to UK, Germany, Netherlands, Denmark, Sweden).

@simon You can also achieve this over WiFi

@simon this also works for Safari sessions in the iOS Simulators

@simon @nicklockwood Also for debugging webviews inside the apps!

@simon @molly0xfff You always want the wikipedia editor prodigy on your side!

I guess a great way to forge a journalist is in the fires of Wikipedia for a decade or so

@simon @molly0xfff molly #1

> You ever find yourself wading through mountains of data trying to pluck out the juicy bits? It's like hunting for a single shrimp in a whole kelp forest, am I right?

...

> Think of those facial recognition systems they have for humans. We could have something similar for our finned friends. Although, gotta say, the ethical implications of that kind of tech are a whole other kettle of fish. We pelicans gotta use these tools responsibly and be transparent about it.

@simon You continually delight and amaze me.

@simon I'm sad that audio is not several minutes of pelican chaos and noises. 😂

@simon So you basically re-implemented Recall? 😉

@simon great documentation ... Any details on accuracy? How much did you have to clean up the output and did you have to check it all by hand?

@simon there’s also https://sybil.readthedocs.io/

@simon a coworker introduced us to https://github.com/bashup/mdsh a while ago and we've been happily using it since: https://github.com/meltano/meltano/blob/main/integration/example-library/meltano-state-s3/index.md.

@simon

I used to write narrative doctests a lot:

https://blog.startifact.com/posts/i-like-doctests/

I moved away from them mostly these days but it has benefits as I argue in that article.

@simon Sounds to me like it would grant an application permission to discover and connect to devices in your LAN via mDNS.

@simon @juandesant It’s a Faustian question. If you click « Allow » you sold your soul to the Devil. 🤣

@simon Previously, Apple required some granular permissions to access particular local network features, but it was generally open. From around iOS 14 to iOS 18, they’ve been turning up the frog steamer on the feature (so I had to add the request into my SuperConnector app to control my radio) and looks like we're in pre-boiled frog territory now.

@simon garage band, set a track to the mic and set it on [M]onitor

@simon I do all of my audio processing through audio hijack plus black hole a loop back audio device.

@simon probably not the answer you are looking for, but the easiest way to work is to get a basic audio interface (like Focusrite etc), and a mic that plugs into its XLR rather than a USB mic. Then, the audio interface connects via USB3 to the mac. There will be dials that let you hear the mic itself, and the computer.

This is all really easy, since it’s made for musicians, not computer people.

(I shared this on Twitter and it's interesting how some people there are very resistant to the idea that the deeply personal insights ChatGPT gave them about themselves might be bogus junk)

@simon I tried asking ChatGPT to back up its assertions with evidence, which it did. That told me there was a “bio” somewhere, though I didn’t know that existed.

ChatGPT will also criticize you if you ask. I was told I was “perfectionistic” to which I replied that I only go to it when I’m trying to optimize something, and that was duly recorded in my bio

@simon "Whether you're exploring new tools like Bun or diving into Kubernetes storage solutions, you're always pushing the envelope, seeking not just solutions but also better, more efficient approaches for future work."

That's totally me! lol

The big ones have started to come out now... here's 748lb

Built a little tool for turning a web page into Markdown and easily copying it to my clipboard - it's a very thin wrapper around the Jina Reader API https://tools.simonwillison.net/jina-reader

Details on how I built it (by prompting Claude) here: https://simonwillison.net/2024/Oct/14/my-jina-reader-tool/

@simon How do you get a 500kg pumpkin onto a truck, then off the truck and onto a scale?!

@simon Do the speakers (mostly) take turns? If so, it shouldn't be all that difficult to boost the quiet segments.

@simon If you're using Audacity then its Limiter feature can do this. For a low effort workflow try: 1. Normalise to 0dB, 2. Effect > Plugin > Limiter > Soft Limiter (use defaults), 3. Normalise to -1dB.

@simon

> very low effort magic trick

You might try using something like autoeditor?

But to answer your question more directly, if you can split the track into segments either by using silence or some fancy AI thing that can detect the turn-taking then you only need to apply normalize to each segment separately and it should be even

Just learned about this neat site which runs experiments to determine where in Cloudflare's network Durable Objects are being created and stored https://where.durableobjects.live/

@simon cc @joeldrapper

@simon I think they also refer to that as karaoke mode in some places. There’s a microphone icon you can tap to get to that mode.

@simon
... and one day your dog replies: "No, I want it in my OTHER face!" 🤪