Discussion about this post

User's avatar
VJ's avatar

Damn I don't know how you do it and to top it off with the writing. All of that is very interesting and I'm tempted to subscribe but at the same time, worried about time organization and being overwhelmed.. and I also have in the back of my mind the opportunity cost of consuming too much information now, if that makes sense.

So on the topic of information overload which segways into your last point of training AI with GPUs--which I take to be sarcasm--there's an interesting alternative, or perhaps complimentary, movement of open source distributed or crowdsourced AI training, model fine tuning and dataset construction. Some articles about this:

https://www.semianalysis.com/p/google-we-have-no-moat-and-neither

https://www.semianalysis.com/p/on-device-ai-double-edged-sword

https://reclaimthenet.org/open-source-ai-is-needed-more-than-ever

As you probably know, there's a lot you can do at home now except it's just slow for one person operation. So there's talk about incorporating cryptocurrency to incentivize a distributed or crowdsourced AI system. This then pertains to the discussion about AI regulation and the increasing fear mongering around it, which I don't buy. There are ulterior motives and negative consequences of such regulation, which would conveniently hamper privacy and any kill wider commercial marketplace offerings, DIY, home lab, independent/crowd sourced efforts:

https://reclaimthenet.org/eric-schmidt-generative-ai-true-anonymity

https://reclaimthenet.org/criticism-of-open-source-ai-regulation-is-based-on-protecting-big-tech

Speaking of which, have you heard of Unstable Diffusion, the *-chan equivalent crowd sourced version built to train it on things that the official Stable Diffusion has not or has been excluded to train on?

To be honest I'm not that enamored with the generative AI systems in and of themselves nor do I subscribe to the doom-and-gloom SkyNet scenario, which only works to provide fuel to the government regulation fire, but I see them as useful tools in certain applications--which machine learning always was, until ML was re-marketed as AI. Any talk of The Singularity is a dead end without first investigating consciousness IMO

You are also the first substacker I've come across who's familiar with old school internet (BBS, IRC) , 4chan et. al. message boards and what Isekai is, AND heath and exercise, and has broad interest in.. well, everything. I will have to get the Barsoom novels now. The oldest SF books I've read related to Mars is Stranger in a Strange Land by Robert A. Heinlein, who's still one of my favorite authors.

Finally, regarding your citation of El Gato Malo's posts here concerning Trojan Horse bills, I have been admittedly spamming different people and places in probably the most inefficient way possible, like I'm doing now, trying to raise the alarm about legislation that also relates to your post, "The Onymity Question". I'm unsuccessful so far, but I'll give it a shot if you don't mind. This legislation, both at the Federal level (proposed) and state level (passed in 2 states) is a direct attack on anonymity/pseudonymity and has worrisome implications not just for privacy, but for security and government abuse. I'll just link to my comment there for all the details:

https://boriquagato.substack.com/p/why-e-verify-is-a-deeply-dangerous/comment/15892825

In a nutshell, the "Protecting Kids on Social Media Act" bill outlaws all means of anonymous mass communications. It requires all "social media" platforms (including the online applications themselves) to identify all users with real world verifiable or government ID. Social media is defined as any medium that allows user-to-user exchange of text, video or audio, either to the public or large groups. Substack will be affected. Even plain ole message boards and large group messaging/chats will be affected -- explicitly called in the linked bill text, see pdf (pg. 4, line 12 - Sec. 6 (C)(x)).

Everyone will have to upload government ID and the people running the platforms and/or regular folks running servers on their own dime will either have to pay commercial verification services to ID all the users or subscribe to the proposed Federal Digital ID service run under the US Secretary of Commerce.

One has to wonder how they'll enforce this for more niche technology that's not centralized.

There's a broader discussion about the Constitutionality:

https://reclaimthenet.org/the-possibly-unconstitutional-nature-of-us-plans-to-force-social-platforms-to-collect-id

Expand full comment
Daniel D's avatar

Wow, quite the comprehensive digest of Substack writing you compiled! (And thanks for listing me in it!) This really is a great platform at the moment for heterodox thinkers to share ideas and insights; I hope and pray it stays that way!

A French press is *almost* the best way to drink coffee. When I was in Iraq, I got introduced to Arabic/Turkish coffee, which is probably brewed with ground coffee and coca leaf. You could wake up a corpse with that stuff! I have tried to make Turkish coffee at home with less-than-satisfactory results -- probably because I'm missing the coca leaf or 8-ball or whatever it was that the Iraqis used.

I thought I was decently well-read, but I see that I am not. I had read some of the posts you referenced, but now I got a shit-ton more great reading material to enjoy going through. I missed the glory days of Social Matter's TWIR, so thanks for filling that role here and now on Substack!

Expand full comment
62 more comments...

No posts