ikesau

I'd watch if you made it. Big fan of pol.is : )

Wow, reading this, I can _feel_ my brain resist an unfamiliar concept. I recommend people read the MDN article linked to in the first sentence before moving onto the examples in the blog.

I'll have to play around with @property myself to get a sense of how it works - knowing esoteric CSS features is a superpower if you've got a complex UI to implement but wanna minimize JS dependencies.

Ah, I love this. Nice work!

I really wish supermarkets were mandated to post this information whenever the price of a particular SKU updated.

The tools that could be built with such information would do amazing things for consumers.

Love p5 - I learned a bunch of new concepts through it: polar coordinate space; thinking in radians; dot products; event-driven programming.

It's one of the fastest ways I know of to make something fun and expressive in code.

Sure, but the fines are progressive, aka proportional to an individual's income.

~10mph over the limit = 75% of your weekly income.

~25mph over the limit = ~150% of your weekly income.

It would not be in the Chinese government's interests to do this in a way that preserves anonymity for its citizens and protects them from being monitored, but I do really want to see a layer of the internet where it can be trusted that every poster is a real human, or a citizen of a particular country.

I understand that this is the sort of thing cryptographers think zero-knowledge proofs could help us with, but I'd be fine with a centralized approach: a government system that provides key pairs and a verification system for its citizens that could be done during the passport application process.

Retrieval-augmented generation (RAG) is a technique for enhancing the accuracy and reliability of generative AI models with facts fetched from external sources

What's the typical curve for a project?

I would have thought it would be like other technology adoption curves - surprised if it _wasn't_ logarithmic

Really good post! I also found this quote which distilled their position in the 404media coverage of the situation.

> “What I can say for sure is that TicketMaster and AXS have had every opportunity to support scam-free third party ticket resale and delivery platforms if they wished: By documenting their ticket QR code cryptography, and by exposing apps and APIs which would allow verification and rotation of ticket secrets,” Conduition told me in an email. “But they intentionally choose not to do so, and then they act all surprised-pikachu when 3rd party resale scams proliferate. They're opting to play legal whack-a-mole with scammers instead of fixing the problem directly with better technology, because they make more money as a resale monopoly than as an open and secure ecosystem.”

from https://www.404media.co/scalpers-are-working-with-hackers-to...

Great, thanks. I'm sure you'll find it thought provoking.

re: hundred million: I was trying to imply that AMCs can be large but not necessarily always in the billions - it depends on the market size and estimated economic damage of the thing we're trying to incentivize R&D for - though yes, both medical AMCs to date (pneumococcal and covax) were in the billions.

Just to clarify, AMCs aren't the same thing as governments "fixing" the price of R&D with wishful thinking. It's when a government/funder uses epidemiology and economics to calculate the cost of leaving a disease untreated, and agreeing in advance to pay that much for it. That way, if set correctly, companies will invest to develop a solution without fear of being punished by the moral/political forces that make them suppress their prices once the solution is developed.

It's a way to fix the incentives that have left us with a whole class of neglected tropical diseases (firms needing to recoup investments, governments needing to negotiate cheapest possible prices)

I highly recommend you listen to the recent 80000hours podcast with Rachel Glennerster on Market Shaping, or read her book, Strong Medicine.

Advanced market commitments, i.e. governments signing contracts that guarantee they'll pay $X hundred million for a solution BEFORE the research has been done, is the way to incentivize this while avoiding the disincentive of governments becoming extremely tough monopsonic price negotiators once the drug has been developed.

The development of more drugs could be funded this way, through global health organizations like GAVI, such that medicine wouldn't need to be developed by gouging some for the benefit of others.

A shame they couldn't control for performance bias. Six sessions with a physiotherapist would probably make a difference in outcomes even without the walking.

Oh nice. I have a static html page I maintain on my website. the simplicity of maintaining that can't be beat, but it would be nice to give other people an easier way of following that if they'd like.

Why is this such a hard problem to solve? is it because of f2p making it trivial for botters to create new accounts once they get banned?

update: pretty exhaustive summary of proposed solutions and their flaws here: https://www.youtube.com/watch?v=SgkgsgaBBCA

looks like it's a firefox (mac, v127) thing. i just tested in chrome and didn't get any issues.

hey sam! were there any other queue types that you decided to omit? and what were the benefits of web components in this article?

just went back and checked! (started running into perf problems by the last exercise)

very cool :)

yeah, i really like the achievement system for the interactive demos. it's a fun way to make sure readers are actually engaging with the lesson, and understanding its points.

in terms of accessibility, i imagine a universal standard for modal state management will be an improvement over the bespoke approach - closing when the user presses escape, focus handling, etc.

but i really wish popover=hint made it into the spec. would have been nice to get a native alternative to tippy.js

There's arbitrary "accuracy lowering" vandalism done by (i suspect) bots that alters dates by a few days/months/years, changes the middle initial of someone, or randomizes the output in an example demonstrating how a cipher works.

it can be hard to spot if no one's watching the article. puts me in a funk whenever I catch it.

I've loved Pat's music for 15 years now. There's an obsessive passion in the people who are into him, trawling for bootlegs and demos on soulseek, binge watching old Women live shows on YT. my friends excitedly recount the email conversations they've had with him, and hell yeah, wavs up for free e-transfer him if you can.

Regardless of whether you like the music, I think the larger object of his art and our reaction to it is terrific and I wish more people could create, discover, and relate to art this way.

as framed in the introduction of this article, https://xkcd.com/1205 is hard to ignore whenever i'm spending time configuring a tool in the name of "productivity"

but i just got nushell properly set up and i've realised that the table doesn't capture the whole dynamic: there's more to work than time trade-offs when new tools enable entirely new types of tasks.

nushell makes working with structured data fun and so i've started benchmarking my code and quickly doing performance analysis - something i was rarely doing before.

you can still ultimately reduce this down to time saved, i suppose, but as a person who's generally skeptical of terminal tweaking as procrastination in disguise, i can highly recommend giving nushell a go.

i don't like cars at all and think it's a terrible shame how they've influenced the design of cities, but i think we might still disagree on what future trends will be :)

if you can entertain the idea that in 10 years there will be far more cars with lidar+cameras+gps around (plus the compute to process it all), steve's point is that this data will be very useful for generating real time road maps such that we won't need to rely on a network of pedestrian volunteers to manually update a database whenever a road gets closed for repair or a new condo is built.

Similarly, as recommended in the article, StreetComplete[1], which shows incomplete areas on your map and marks them as quests to add to :)

Also recommended, Organic Maps[2], which is an OSM client that's good enough to use as my main map app (in my city, at least)

Final recommendation, this interview[3] with the founder of OpenStreetMap Steve Coast. 3 hours of fascinating insight into the history and future of the project, including how he doesn't think OSM will be relevant for much longer once someone manages to develop a mapping project into their self-driving cars

[1] https://streetcomplete.app/?lang=en [2] https://organicmaps.app/ [3] https://www.youtube.com/watch?v=uB9NxbnE1i0

it's like.. the hacker wouldn't state their ulterior motive so plainly, but they did and Nicholas still fell for it.

I think it's good to examine all habits that we have, and if this article gets you to consider the (potentially unknown) trade offs you're making, great!

but to really know what they are, RCT self-experiments are the best way to work out what's really going on.

gwern has a bunch of examples (https://gwern.net/zeo/caffeine) but you should tailor yours to the effects you're most interested in.

incidentally, does anyone here have a go-to tutorial for how to set one up, how to do power analysis, etc? i'm persuaded by why we should do RCTs, but I don't have the background in statistics to confidently run them myself.

i'm not sure i understand you. you're saying third party app stores will incentivise apple to improve their quality control to better differentiate them from competing app stores?

i recently saw a refreshing perspective on canadian immigration that addressed these issues at their origins: failures in housing policy and complacent governance. it's the perspective i wish more canadians held so we could steer the government towards policy that's better for all. maybe nativist nimbyism isn't inevitable.

https://www.youtube.com/watch?v=hQkzE9PsAv4

> Luis Fernando Ramadon, a federal police specialist in Brazil who studies extractive industries, estimates that the global illegal sand trade ranges from $200 billion to $350 billion a year—more than illegal logging, gold mining and fishing combined. Buyers rarely check the provenance of sand; legal and black market sand look identical. Illegal mining rarely draws heat from law enforcement because it looks like legitimate mining—trucks, backhoes and shovels—there's no property owner lodging complaints, and officials may be profiting. For crime syndicates, it's easy money.

> The environmental impacts are substantial. Dredging rivers destroys estuaries and habitats and exacerbates flooding. Scraping coastal ecosystems churns up vegetation, soil and seabeds and disrupts marine life. In some countries, illegal mining makes up a large portion of the total activity, and its environmental impacts are often worse than those of legitimate operators, Beiser says, all to build cities on the cheap.

this sounds like a really hard problem to fix. i see there are researchers trying to estimate mining levels by counting ships[1], but even if we're able to get a heatmap of the problem, there'd need to be an enormous amount of cooperation between nations to certify sand provenance. damn.

https://eos.org/articles/satellites-spy-on-sand-mining-in-th...

Cloze cards also work well for this sort of thing, I think. Though they're easy to mass produce and create an overwhelming amount of revision, so be careful.

You may not have noticed, but there's also a reverse card type that only requires one note but adds two cards to your deck.

There are lots of people who have a similar intuition on the ChatGPT thing but none that I've seen create cards as good as the ones you could make yourself. Using ChatGPT as a dynamic pen pal is probably the better way to incorporate it into your language learning.