634636346

Fitting that this was posted on Columbus Day.

> Cathie Wood’s Ark ETF seems to find the payments margin opportunity, and the gang to get it there. In its Big Ideas 2023 paper, Ark points out that there are 9 steps between Buyer and Seller in a consumer payment transaction, sucking intermediary fees of roughly 2.8% of the value of the purchase. Ark believes enormous money is to be made if the steps between Buyer and Seller would be reduced from 9 to 3 (removing card networks, issuers, and acquirers, for the most part), thereby reducing the expense from 1.64% to 0.21% of the value of the transaction (leaving a massive 2.60% ‘take rate’ for its horse). If you’re looking for crazy, Cathie Wood’s group is a hot mess full of them.

Most of that "take rate" goes to the issuing bank, which returns it to the customer in the form of rewards, which can often be redeemed for cash, and a number of issuers now offer generic 2% cashback cards. There's also the added benefit of fraud protection (and the ability to charge-back, as a last resort). Anyone using Cash App or similar in lieu of a credit card, I just assume is financially illiterate.

> However, OpenAI pinky promises they don't use API data for anything, like training. Maybe that makes you feel a bit safer, although probably it shouldn't.

It doesn't. I don't trust OpenAI or Sama. Frankly, I'm even hesitant to use VSCode now, even with its customizable privacy/telemetry settings (though I can at least limit its network access).

Doesn't this give OpenAI access to your source code?

It's not just hypocrisy. Many of the HNers railing against this are actually responsible, even if just indirectly, for it. Take Chrome (including Chromium) use, for example. Chrome has never been that much better (e.g. 2-3x) than FF. Maybe at its best, when it first debuted and V8 wowed everyone, it was 20-30% better--not enough to justify the investment a poweruser (who heavily customizes their browser) would have to make to jump ship, and not enough for anyone concerned about the open web. Yet I would guess most people here jumped ship at some point (probably when it was new and shinny and Google still paid lip service to "don't be evil"), and they've never looked back, despite FF having caught up and remaining competitive.

Asking people to not use Chrome isn't asking much, and yet people here can't even manage that.

> engage in illegal

Well then why haven't they been prosecuted?

I wouldn't be surprised if the author, and a large segment of HNers agreeing with her, did a swift about-face when they realized that Tor also provides an end-run around the internet backbone black-holing of IPs that some Tier 1 ISPs did to KiwiFarms last year, during the height of the campaign to deplatform it. More people using Tor in general means more people having the means and know-how to evade censorship, and we can't have that, can we?

I wonder how many HNers angry about this are also some combination of 1) working for bigtech 2) using iOS/OSX/Android instead of Linux (yeah, I know Android is technically a Linux) 3) using Chrome/Safari instead of Firefox and 4) have endorsed, at least in the past, bigtech firms like Google and Cloudflare acting as arbiters of what is/is not acceptable content for the internet, and even whether it should be viewable by anyone at all.

Like one of the commenters in that thread said, this sounds like they were using the noidex feature to use the IA as a personal private backup, and thus abusing it, and ruined it for everyone else. The IA is great as a personal public backup. (For example, I've deliberately submitted copies of certain OSS projects I've worked on to the Wayback Machine.)

Sorry, I wouldn't have been so harsh had I not suspected the entire blog was AI spam. Good luck with your back. (And please do at least some basic resistance training, even just once a week.)

> The text content of the blog is all hand-written by me, not AI, although I do use AI (Stability) to create the header images.

Well, you can't really fault people for suspecting you're using AI for other things too, can you? It's almost a form of gaslighting. You also apparently did an AI-generated podcast? (According to your profile.)

Theo isn't a "Code of Conduct" type of guy. Not a good fit. Though the license of the Rust compiler (MIT/Apache) does at least make it a possibility.

I stand corrected. However, I still think he's using GPT to generate much of his content, and at most doing some post-hoc editing of it.