[brotherdarul]

Well, hello again.  After a lapse of (years) I’ve re-subscribed here mainly to support Mr. Roemmle’s cool garage projects, like YourAI and Grokpod.  I’m still @brotherdarul on X.  See y’all around.

[brotherdarul]

Linnea,

Second the idea of discussing Bitcoin layer 2.  I’m trying as a layman/small investor to assess where cryptocurrency technology is going, and whether the market will favor a “good enough” solution like grafting a Layer 2 onto Bitcoin’s blockchain to adapt Bitcoin to other purposes outside of “Satoshi’s” design, or will Bitcoin remain mainly a (legacy) store of value, with most transaction activity shifting to a blockchain like Ethereum which was designed from its beginning to do so many things natively that Bitcoin’s blockchain was never designed for.  Ethereum has seemed to be very nimble, shifting from its initial proof-of-work model to proof-of-stake, and it’s my impression that several (most? all?) of the Bitcoin layer 2’s do something similar, loading a quantity of Bitcoin onto their particular side chain as a stake, doing whatever transaction work, and then later publishing the result of all those transactions back to the Bitcoin blockchain.

So the question for me is, whether its really going to end up being a choice between Bitcoin versus Ethereum (as I’d originally thought, with Bitcoin ending up as legacy technology, like the post office), but more like a question of which Bitcoin Layer 2 is going to dominate, or is Ethereum going to win out over these Bitcoin Layer 2 projects and end up taking on that role as merely one of its features.  I’d initially thought of the crypto market in a “which standard will win out/network lock-in/VHSvBetamax” paradigm, but because it’s ultimately all software, I’m now doubting that one will eventually outcompete the other, and we’re probably going to end up with Bitcoin’s blockchain being a layer of essential legacy code that someone needs to maintain, that’s a foundational layer on which a lot of other stuff that works better is built.  That this economic technology is being developed alongside AI is going to make for a very interesting future.

[brotherdarul]

Oh, I see that what I just posted in “Read Multiplex Member Community” actually belongs in this category…

https://readmultiplex.com/forums/topic/regulating-malicious-ai-deep-fakes/

[brotherdarul]

Alas, the link turned to HTML tag salad.

My apologies.

The Substack is in my Multiplex profile.

[brotherdarul]

Hi, I’m an attorney and writer from Memphis, TN, and have a Substack blog that covers technology, politics and law.  My most recent article, published this morning, discusses Google’s AlphaFold, a bit about the protein folding problem in molecular biology, and hopefully provides a general interest survey of various applications for transformer AI models.  I welcome your comments!
<div class=”status__content status__content–with-action” tabindex=”0″>

Parlez-vous Protéines?
by Dar ul Harb, Esq.

<span class=”invisible”>https://</span><span class=”ellipsis”>darulharb.substack.com/p/parle</span><span class=”invisible”>z-vous-proteines</span>

</div>
<div class=”status-card__image”></div>

[brotherdarul]

Interesting paper.  Near as I can understand, the subtle form of the poisoning attack creates something analogous to a conceptual “double exposure” by processing the poisoned training data in a way that combines the trained concept with an “anti-concept,” presumably hidden in variations of the poisoned training image’s pixels in a way that doesn’t obviously disturb the image’s correlation to the trained concept, on a per-image basis.  But once a diffusion model has seen enough of these subtly poisoned images, the “anti-concept” present in each can outcompete the trained concept.

[Author]

Posts