wanderingbit

Thank you! I have been reading these QLoRa posts in the hopes of training it on my notes stored in Notion, but then you do it for me! Nice product ;).

I went in to this article hoping for some evidence on why harm reduction isn't always the right choice. Instead it is tone-deaf (as if fentanyl users and heroin users are in the same category as those who choose to use alcohol or pot!) and filled with correlation-implies-causation assertions.

If anyone has some credible evidence for where harm reduction doesn't work, please reply!

Problem (1) can likely be addressed by restricting the number of iterations to some small constant, like 3 to 6 (which they suggest in the article). You can also restrict the number of peers-of-peers you fetch during each iteration to a small random subset. So if we only iterate 6 times and choose 10 peers for each iterations, we’ll get 1 million (10^6) trust scores needing to be pulled to calculate a trust score. That is an upper bound and will likely be less because it assumes each peer is distinct. At 32-bit floating points, that’s 32 million bits or 4 MB necessary to be fetched.

I can imagine this being reduced by at least a factor of 10 without much impact on the trust score. But note the “random subset” means different people will have different trust scores for the same peer :shrug:

For (2) yeah we probably need to lower our expectations on privacy for the time being; it’s a masters thesis and privacy in open distributed systems is very tricky.

It is detailed comments like this that make me feel very fortunate for the community here. Thank you!

Maybe one day there will exist a happy medium between centrally-planned Zerg economy and a free “Ebay for organs” market.

That’s now how I understood the quote.

I thought of it more like if you went to go hire a doctor, then the best people to review that doctor would not be the university that gave the doctor their credential, or even the doctor’s patients; it would be the doctor’s colleagues who understand the doctor’s strengths and weaknesses best because they are the only ones who have the direct relationship as well as technical skills to compare against.

I think at the end of the day we want both perspectives. We all benefit from having people from both inside and outside the system critique the system. Certainly, the person who walks the walk and talks the talk deserves more street cred, but from the my layperson’s perspective I’d much rather live in a world where I can read Linkola’s and Illich’s critiques and take the best parts from both of them.

Your comment primed me for some wikipedia shenanigans, but when I looked up and translated the 1919 original, it’s more or less exactly the same as the Wikipedia summary.

http://web.tiscalinet.it/regno76/testi/manifesti/Il%20manife...

Thanks for prodding me to go down that rabbit hole.

Seems like Marinetti didn’t foresee how horrible the “industry councils” would be.

The nationalized military industrial complex could or could not be a good idea.

This was a great in-depth exploration of some of the ideas from Banks' books that I thought deserved a deeper drive. Even though he was writing in the 80's Banks has a lot of foresight into how modern tech such as AI could result in a society like The Culture. It's definitely utopian, but to the author of this blog post's credit it comes with a subtle lack of meaning that some would find unbearable.

Reading this blog post gave me insight into our my own (Western) culture, and modernity. What happens as we move away from ascribed identities, like be born and dying a feudal serf, to achieved identities garnered from the actions we choose in our lives? I laughed at the jibe about the 2 ways a modern person finds meaning; either embracing their Celtic heritage or going full memeplex!

The website design is unreadable, except if you highlight sections of the text which make it straightforward to read. No idea how you'd read this on mobile, though.

> combat is once again dominated by the kind of melee encounters that were also historically the domain of a trained warrior caste

Correct me if I'm wrong here, but does Herbert ever give a reason for why the characters in Dune do not use gunpowder/projectile weapons (a.k.a. guns), similar to modern armies? I never understood why the characters in Dune bothered with this melee warfare, except maybe for ceremonial reasons.

It's worth reading the whole article. If you've ever read any of Iain Bank's books you'll definitely appreciate it.

I’m not sure why the academics are discouraged , but there are non-technical reasons to discourage you (and all of us) from trying to use Internet-based voting systems for important elections:

https://engineering.stanford.edu/magazine/article/david-dill...

This article voices the general opinion I’ve read from technologists and those who’ve studied networked voting systems. The problems all boil down to any software-based (i.e. non-paper-based) voting system will almost certainly contain bugs, and this will allow elections to be influenced by a few persistent hackers.

Maybe this is where blockchain-based systems could help push such systems forward, by effectively giving a worldwide bounty for voting software. After many years and 0 hacks, we could be pretty confident that this particular piece of solidity code is solid and could be used as part of an Internet voting scheme.

Though, the moment you have a frontend (which you will of course need to interact with the smart contract) you’re back to square one because of all the millions of lines of code (uh oh, bugs!) that are necessary to be executed in order to display a form on your screen.

Maybe it’s a good thing you were discouraged from this bleak area of research, who knows. I’m sorry you were discouraged like that.

There was a recent post about an open source tool for indexing Discord content and making it available for Google search:

https://news.ycombinator.com/item?id=36383773

Me too! I am trying to build a Discord-based remote course and am excited to read through the code here and see if it matches my needs, or can be tweaked to so.

Once I do that I'd like to DM you with some questions mid-kid.

Nice job on getting so much implemented and open for users!

Is this using steganography to create un-tamperable watermarks (which would allow downstream users of possibly-AI-created material to prove that it is indeed created by an AI) or, is this for something different?

To those who dismiss with "correlation is not causation", you should share with them Haidt's in-depth article showing multiple experimental studies regarding social media and mental health, and how more social media use causes lower mental health:

https://jonathanhaidt.substack.com/p/social-media-mental-ill...

Key paragraph is:

"In sum, we found six quasi-experiments that looked at real-world outcomes in real-world settings when the arrival of Facebook or high-speed internet created large and sudden emergent network effects. All six found that when social life moves rapidly online, mental health declines, especially for girls. Not one study failed to find a harmful effect."

That being said, I think relying on a preponderance of evidence is acceptable in this case where the current status quo (particularly for teenage girls) is worrying for the vast majority of people in your community.

There isn't a good excuse to wait. There's no strong evidence that technology can "get us out of this".

Upon reading the OP's blog, I had the same idea! It's cool to see others are hungry for this as well.

My basic approach to implementing it in Notion (my preferred knowledge manager) is:

1. build some block template in notion that lets me structure the block such that it’s easy for Mochi to ingest it 2. reads new pages from Notion 3. uses Mochi to generate cards for them 4. build a habit of reviewing Mochi cards, then I will actually learn from the things I write in Notion

It looks like you've implemented the core functionality in retaind. How come you decided to go with a separate knowledge base, rather than try to use one that already has strong network effects, like Notion?

For those discussing what it looks like for a polity to have a widespread ban on types of certain types of advertising, look no further than the great state of Vermont!

We have not had billboards for as long as I've been alive, and it is great, especially when you head over to NY or Massachusetts and see how tawdry all the billboards look.

Read more about the campaign to ban billboards and Ted Riehle here: https://vtdigger.org/2021/07/11/then-again-vermonts-billboar...

Context: I've been working in the cryptocurrency space for a decade.

One of the most interesting technologies the crypto space is working on are the SSI (self-sovereign identities).

https://en.wikipedia.org/wiki/Self-sovereign_identity

Even if the whole idea of cryptocurrencies is found to be a dead-end, the fact that it invigorated the research and development of SSI will change some important things about how we operate online.

There exist prototypes of tech that allows you to prove you are indeed a unique human being online [1], and reveal nothing else about your identity. Most importantly, this tech is not owned or controlled by any FAANG or government, it's an open protocol just like email.

I have listened to a podcast with an expert researcher in AI, and I remember him saying that he predicts some form of cryptographic identity will arise in order to help deal with the bot problem

[1] https://worldcoin.org/ (note, I don't work for them, and have no idea if this will be the tech that finally breaks out, I just think they're the furthest along of any of their competitors)