Hacker Newsnew | past | comments | ask | show | jobs | submit | arbuge's commentslogin

> In a few years, there will be hardware capable of running frontier models good enough for most things at accessible prices for even tiny companies.

What makes you so confident about this prediction? Hardware costs haven't exactly been cratering recently.


.> Hardware costs haven't exactly been cratering recently.

No, but local models have been booming in performance/quality improvements. The RAM shortage won't last forever (more supply will come online when if demand doesn't diminish), and then the math would be pretty easy.


Can they actually make money at these prices?

Why do you think they have to?

The EVs I'd really like to see some day in the US are the dirt cheap 100 mile range $5k Chinese EVs.

One can dream I guess.


I dunno. While most people likely only need one of those, it just feeds into the narrative that you can't take EVs on road trips.

Likely to remain a dream as the people buying a $5k car are likely to prefer a reliable 12 year old luxuery car instead.

They don't need US crash or safety standards. Just get an e-bike instead.

> free-to-use pay phones

Some redundant words there perhaps.


Why? Payphones have never been distinguished by the fact that you had to pay to use them. You also had to pay for the phone in your home.

Payphones were distinguished by the fact that they were located in convenient public places, and if you needed to contact someone, you could use them. That's still true here.


You could use a pay phone to call the operator. You can use it to make collect calls. Hell, if you were industrious enough, you could trick the phone into giving you a dial tone for free. The VoIP ones will probably be harder to trick though

People know what a payphone is, and the service it provides. As he said in the article, he chose the payphone style for what it signals to people - that it is public infrastructure.

As it pertains to AI, I think we will eventually come around to the conclusion that consciousness is not a useful construct.


> Claude Design empowers non-designers to make decent designs. It’s not aimed at designers.

Well, when you put it that way, that sounds bad for designers, and, by extension, Figma.

ps. I do like commas.


As a proponent of the Oxford comma, I didn’t mind those commas.


Those are not Oxford commas, they’re parenthetical (and I like them too!)


I know him from Harvard and came here to say pretty much the same thing. RIP.


I took his Computability class in the Hebrew University. He got angry that students were often late to class, and said that this never happened in Harvard...


Any of his "chalkboard" lectures (preferably in English) in open archives of these universities? YouTube searches only bring up Prof Rabin's lectures aided by slides and presentation (ex: https://www.youtube.com/watch?v=thK_qJqx5mo at Tel Aviv Uni / https://www.youtube.com/watch?v=QCX0Ut0hcWw at Harvard).


There's a chalkboard here (used ~44:50):

Cryptography and Preventing Collusion in Second Price (Vickery) Auctions - Michael Rabin

https://www.youtube.com/watch?v=4cmCBVrVQqc

No chalkboard but more lectures

https://youtu.be/nbePExzSTQ0?si=KkTbwfwj5rMtQUhD&t=681 - פלאי תורת ההצפנות ויישומיה לתהליכים פיננסיים (The wonders of cryptography and financial applications)

https://www.youtube.com/watch?v=N_LG5Hcc8mM - Lecture 7 - Zero Knowledge Proofs and Applications Michael Rabin

For those interested in searching for more here's a Hebrew search string you can use: "פרופסור מיכאל רבין הרצאה" interesting enough Google and YT search yield results in English and Hebrew but possibly different ones than just searching in English.

EDIT: One more:

https://www.youtube.com/watch?v=30wkb46BE1k


Thank you. Appreciate it


They state the link is claude.ai/design, which currently goes to:

Page not found Claude can help with many things, but finding this page isn’t one of them.

when logged in.


I'm far from a crypto expert but aren't costs largely GPUs and electricity here?

Those are now being driven by massive AI demand and are likely to remain so for the forseeable future. So how would costs go down?


The cost of finding a block goes down because it becomes less difficult.

The goal in proof of work is to find a block hash less than a given value. That value is determined by the network difficulty. The lower the value, the more difficult it is to find a block, and thus the more expensive it will be to mine.

Difficulty is adjusted once every two weeks to target an average block time of 10 minutes. If the average block time during the preceding 2 weeks is less than 10 minutes, it means that blocks were too easy to find (i.e. the difficulty was too low relative to total hash rate of the network). Conversely, if the average block time was greater than 10 minutes, the difficulty was too great.

This is how it the network has maintained a roughly 10 minute block time as the hash rate of the network has grown over the past 16 years. The difficulty (i.e. cost) of finding a block is constantly being adjusted.


I don't think GPUs are competitive at all. You need specialized mining rigs with bitcoin mining specialized chips.


And that since a solid decade.


Bitcoin is no longer mined by GPUs but by ASICs


Don't the ASICs compete with the same fab capacity that fabs GPUs, RAM, SSDs etc.


You’re fractionally right with GPUs but RAM and SSDs run on different processes at different fabs.


They compete with older GPUs. Not new ones, not RAM, and not SSDs.


If costs stay high, then people will drop out of bitcoin mining, which will cause supply to go down and bitcoin prices to go up.


It won’t cause supply to go down, the same amount of Bitcoin is produced whether it’s mined by millions of ASICs or a single 2008-vintage laptop.


Perhaps they'll pro-rate it by size.


Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: