strangecasts

> It seems like a decent amount of the evidence regarding puberty blockers was excluded for dubious reasons.

To be precise, the report excluded studies of blockers for not being blinded, in a recreation of the 2003 review of parachute efficiacy [1]

Not that it's particularly consistent about any evidentiary standard, as they happily cite a "self-identified sample of 100 completing an anonymous online questionnaire" later on

[1] https://www.ncbi.nlm.nih.gov/pmc/articles/PMC300808/

Others have been able to provoke the same prompt, so unfortunately not a joke

According to the FAQ the current implementation is in C#:

> Second, you might ask: why write it in C# first?

I interpret their strategy as sticking to a subset of C# which works with NativeAOT -- for compilation to native libraries -- while also working in the .NET runtimes shipped with Godot and Unity (is IL2CPP still around?)

Thank you for providing an exhaustive list of references :)

> Finally, specific to XAI, we're seeing people actually deal with the problem that, well, people aren't really using this stuff!

I am very curious to see which practical interpretability/explainability requirements enter into regulations - on one hand it's hard to imagine a one-size fits all approach, especially for applications incorporating LLMs, but Bordt et al. [1] demonstrate that you can provoke arbitrary feature attributions for a prediction if you can choose post-hoc explanations and parameters freely, making a case that it can't _just_ be left to the model developers either

[1] "Post-Hoc Explanations Fail to Achieve their Purpose in Adversarial Contexts", Bordt et al. 2022, https://dl.acm.org/doi/10.1145/3531146.3533153

Article from 2022 - I am not sure how far it is along in development?

Super interesting!

Immediately reminded of Iterative alpha-(de)Blending [1] which also sets out to set up a conceptually simpler diffusion model, and also arrives at formulating it as an approximate iterative projection process - I think this approach allows for more interesting experiments like the denoiser error analysis, though.

[1] https://arxiv.org/pdf/2305.03486.pdf

Reminded of Fate of the World [1], a similar take on the concept from 2011:

[1] https://store.steampowered.com/app/80200/Fate_of_the_World/

Unfortunately the download is taking its time - which kind of base model is it using and what techniques (if any) are they using to offload weights?

Since the demo is 35 GB, my first assumption was it's bundling a ~13B parameter model, but if the requirement is 8 GB VRAM, I assume they're either doing quantization on the user's end or offloading part of the model to the CPU.

(I also hope that Windows 11 is a suggested and not a hard requirement)

Steam does have Curator Connect, which kind of does that - https://partner.steamgames.com/doc/marketing/curators - granted, it comes with a couple of drawbacks, the main one being that the reviewers need to have a Curator group set up

Still genuinely unsure of how they got FDA approval for human trials close to a federal investigation into their treatment of animals [1], I have not done human subjects research but I cannot imagine an EC going "oh well carry on then"

[1] https://www.forbes.com/sites/siladityaray/2022/12/06/musks-n...

Admittedly not familiar with Mobian - I think the biggest differences are that postmarketOS is based on Alpine Linux (and musl in turn) and encourages porting to a wide variety of devices, while Mobian builds on Debian and attempts to provide more complete support for a smaller number of phones

Really glad to see PostmarketOS becoming more viable as a daily driver - there are so many devices which should still work well, if not for Google making OS updates dependent on SoC vendors and carriers...

(e: yes I know there's Treble now but you'd really hope "multiple SPoFs who would all rather you bought the new phone" came up earlier as an issue with updating)

As mentioned in the article, Andreessen's "techno-optimist manifesto" cited Marinetti - one of the coauthors of the original fascist manifesto - as one of his inspirations

I found it strange that one of the stated reasons behind the verdict was that the court believed that Nature and Youth's alternate proposal [1] of storing the waste in underground depots would not be any more environmentally friendly than the alternative [2]:

> "Viktigst er likevel at tingretten ikke finner mest sannsynlig at et driftsalternativ med lagring av avgangsmasser i dedikerte bergrom vil være miljømessig vesentlig bedre enn sjødeponi – alle miljøvirkninger tatt i betraktning." ("The most important fact is that the court does not find it likely that mining operations with storage of mining waste in dedicated subterranean depots would be a substantial improvement environmentally on storing it at sea - with all environmental consequences factored in.")

[1] https://drive.google.com/file/d/10bcrk0Iyvn8eouBshKcNGxe35fc...

[2] https://www.tu.no/artikler/nederlag-for-miljobevegelsen-ting...

> (if this is what really is happening)

TU explains the process [1] - the waste is mixed with seawater and a flocculating agent, but it is fundamentally still being dumped

[1] https://www.tu.no/artikler/nederlag-for-miljobevegelsen-ting...

Especially frustrating, since, as the article points out, Norway is party to the Aarhus convention which specifically requires that environmental cases should not be "prohibitively expensive" ("uoverkommelig dyre"), but the government excluded "side costs" such as lawyers' fees and expert witnesses from their estimated cost of a trial [1]

[1] https://issuu.com/deninternasjonalejuristkommisjon-norge/doc...

> before the company even acted on what we sent them, Substack shared the scope of our findings with another, friendlier publication on the platform

Everything else aside, if they can't help themselves from doing petty stuff like this, why would you let them manage your subscriptions?

Really cool! How did you translate the source code to TypeScript?

Demoed by Sonicstate: https://www.youtube.com/watch?v=Q5Y7GQzIatg

> In November 1967 the Israeli authorities issued Military Order 158, which stated that Palestinians could not construct any new water installation without first obtaining a permit from the Israeli army. Since then, the extraction of water from any new source or the development of any new water infrastructure would require permits from Israel, which are near impossible to obtain. [...] They are unable to drill new water wells, install pumps or deepen existing wells, in addition to being denied access to the Jordan River and fresh water springs.

https://www.amnesty.org/en/latest/campaigns/2017/11/the-occu...

I was thinking more for prototyping there ;)

(I'm assuming at this point you are trying to set up a first prototype or making something you can assemble yourself in a small run)

I think that might be an error in the product photo- the 61TP/9 would have 5 octaves. These would also be full-size keys, I'm not sure if there are OEM kits for smaller keys - maybe Microkorg or Minilogue replacement keybeds could be good for that?

You are correct that the keybed will probably be the expensive and tricky part of this - lots of DIY synths and controllers opt for push-button keyboards or capacitive touch for that reason. For small scale prototyping, there are templates for 3D-printed keys [1] and Doepfer also offers OEM kits for piano keys [2] (with unweighted keys being 175€ and weighted ones being 600€).

[1] https://www.thingiverse.com/thing:3760403

[2] https://gael.doepfer.eu/index.php/en/item/manual-fatar-61tp9

The technical issues did manifest, e.g. the 600 tweet per day limit [1]

[1] https://www.engadget.com/twitter-puts-strict-cap-on-how-many...

There is also the Reface DX Legacy project which tries to convert FM synth presets - including the TX81Z presets - to work with the Reface DX: https://refacedx.martintarenskeen.nl/

You don't have the full set of waveforms, no, but IIRC the operator feedback option can be used to get square and sawtooth waveforms as well

For fields where ArXiv preprints are the norm, Arxiv-Sanity [1] allows filtering by keywords and basic topic filtering.

In ML, I personally like skimming Davis Blalock's Davis Summarizes Papers [2].

[1] https://arxiv-sanity-lite.com/

[2] https://dblalock.substack.com/

This is how I view it as well, if you're working on a library/framework which is an inextricable part of other people's products, you have a moral obligation not to participate in enshittification, in the same way you would expect other people to say, not inject ads into your browser addons

The problem, on multiple levels, remains rewarding people for doing necessary maintenance work rather than "big-ticket" features like adding another ad framework

Many people do work on their own engines, and do not have a game to show for it - which isn't to say they aren't cool projects, but it does involve reinventing many things people have already done, and also takes away time that could be spent prototyping

To add to the big pile of options, HaxeFlixel [1] and Defold [2] come to mind as lighter 2D-focused frameworks with web export options and published games

[1] https://haxeflixel.com/ [2] https://defold.com/