Alex3917

> I think it would be totally incoherent to say that copyrighted information is essentially fair game to include in your model but the outputs of your model are privileged against being included in other models.

IANAL, but going from copyrighted texts to an AI model likely constitutes a new original creation, not a slavish reproduction. Whereas going from one AI model to another is more likely to be considered a slavish reproduction reproduction, although arguably AI models are not copyrightable to the extent that the weights are objective facts, like entries in a phone book.

> That’s the part that defies common sense.

Does it? Isn't this basically the same as how cellular companies can figure out exactly where someone is calling from as long as their phone is pinging 3+ towers? Just connect the towers into a box and you have the same result.

> nobody is claiming prediction markets are a type of journalism.

James Surowiecki wrote an entire book essentially saying otherwise.

For converting emails into archival formats, the two open source tools I'd look into are ePadd and Mailbagit.

If at least some subset of these emails are things you'd like to explore making searchable/viewable by the general public, I run fwdeveryone.com and am happy to chat.

Yes, that was a political decision based on the fact that people with Asperger’s weren’t able to access sufficient social support. But from a medical perspective, the two have basically nothing to do with each other; Asperger’s is primarily about the connectivity between brain regions, whereas (traditional) autism is more about functionality within brain regions. Obviously there is a lot of overlap and this isn’t necessarily the mainstream scientific interpretation, but I think it’s the most parsimonious way to define these terms.

I mean you can literally upload your genetic methylation variants and bloodwork into Gemini and have it one shot a custom treatment plan for you, how much more personalized do you want? Obviously if you want hand holding from the medical system you can pay a psychiatrist or a functional doctor to do the same thing, but at this point the idea that treatment isn't personalized enough is frankly absurd.

My point is that autism is already treated by the medical system as many disorders and not as a single thing. The reason they all share a common label is for political reasons, where it makes sense to group them together. But the actual medical system does not treat them as having anything in common, let alone as being homogeneous.

> what leads us to treat ASD as an actual single disorder, as opposed to a billion different individual neurological profiles?

We already do treat it as the latter. The medical definition of autism is basically just any neurodevelopmental disorder that isn’t ADHD.

North American waters also. The waters off the coast of Cape Cod are being decimated by Chinese-funded vessels, based on what I've heard from local fishermen there at least.

It's going to be replicated because we already know the mechanism of action. That said, many/most people would probably be better off solving the problem more upstream via whatever B-complex vitamins they need based on their bloodwork and DNA methylation variants.

> the reality is that engineers are pumping out code faster than the humans can understand and (honestly) review it

The word "engineering" means doing things using a repeatable process to get predictable results. If you can get predictable results (e.g. guaranteeing the absence of data integrity issues, security issues, anything that will cause downtime, etc.) without looking at the code, you're still doing the work of engineering.

Your job as an engineer is to choose high-value problems to work on, solve them in the correct way, and guarantee that the functional and non-functional requirements are met. If you no longer need to read the code to guarantee that it has the right set of ilities, then I'm not sure that's obviously a bad thing, as long as the ilities you're enforcing result in a codebase that is sustainably secure and maintainable over the long term.

> I've lately been wondering if schizophrenia causes this connection-seeing, or if connections cause indicators of schizophrenia/psychosis.

IANAD, but my reading of the literature is that the genes for schizophrenia cause connection seeing, which on its own is mostly net neutral (slightly negative on average, but positive for many people). But it’s not having the ability to think critically about those connections, along with trauma and lack of social support, that causes this ability to see connections to manifest as mental illness. The genes for schizophrenia on their own though are, on average, only slightly net negative in terms of their impact.

I think the parent comment was a reference to the famous “Economists must learn to subtract” commercial that AdBusters ran.

> checking any text for watermarks requires sending the entire text to Anthropic

Couldn’t it be checked in the TEE using confidential computing to keep Anthropic’s algorithm secret?

Really? I still have my 9.95 PayPal email receipt from buying the game in 2010, but I don’t have a single email sent to that email address about any sort of migration.

Email me and I can connect you with some. There is a FB group for folks growing them in the NYC area, plus a list of all the parks where they are planted.

> Homeopathy was invented after the discovery of germ theory [1][2]. So not really.

Germ Theory was only finally accepted (after initially being rejected) due to the advent of evidence-based medicine, which homeopathy popularized.

There is a huge amount of overlap between the genes associated with schizophrenia and the genes associated with autism. This is part of what makes it tricky to diagnose yourself, because anyone who has the genes for schizophrenia is also going to have a lot of autism symptoms even if they don't have autism.

Not to be confused with Google’s existing product called Wiz.

Having worked at both greenfield startups and unicorns, I've found that virtually every problem I've encountered at the unicorn startups was caused by folks being incompetent at the greenfield level. Maybe when you get to the scale of Google things are different, but it's certainly possible to build a business big enough to retire off that doesn't require any more technical knowledge than what you'd learn at a two-person pre-PMF startup.

For both Amazon and Nvidia, their marginal costs are probably much lower than their fixed costs.

> Did they buy some code for Gmail from an older company?

They bought both Deja and Neotonic.

I mean the I would really only include the code for things like:

- Fetching email messages

- Parsing email headers

- Mime parsing

- Converting the text of email bodies into UTF-8

- Threading messages

- Eliding reply text

Given that the official story is that pb made the first version of Gmail in a day, does anyone actually believe that he wrote the code for any of those things in a day? If you honestly believe that I have a bridge to sell you.

Wait till you learn that the source code in Chrome also predates the existence of Google.

This is a pretty good talk on the history of email: https://www.youtube.com/watch?v=mrGfahzt-4Q

He wasn't sitting there writing binary code and implementing all 7 layers of the OSI stack by hand, he was was gluing together pre-existing components. And the pre-existing components he had access to include two major email startups acquired by Google in 2001 and 2003, which were founded in 1995 and 1997 respectively. (Although he does have at least two patents for features and algorithms he co-invented while making Gmail.)

> You guys cloned a whole suite of products in a short period of time that cost millions of dollars.

At the risk of stating the obvious, the functionality isn't actually cloned, only the UI. The actual code powering Gmail probably dates back to the late 80s or early 90s and has had several hundred thousands of hours of work put into it. This is just a webpage that looks kind of similar.

I point this out only because I've seen people saying that software businesses don't have moats anymore because of this, which is taking away a completely false lesson.

> Prompt engineering is like a shittier verson of writing a VBA app inside Excel or Access.

Sure, if you could use VBA to read a patient's current complaint, vitals, and medical history, look up all the relevant research on Google Scholar, and then output a recommended course of treatment.

This is the key issue. There is zero doubt whatsoever that flossing is essential, and the fact that the empirical evidence is equivocal shows the limitations of science to prove even the most obvious things.

> What’s stopping a similar crisis that 23andMe customers faced where their genetic data along with their identifying information getting sold to the highest bidder if you ever become insolvent?

Nucleus employee here. Nucleus is a medical provider that is providing a medical service and is regulated by medical laws, which extend even through bankruptcy or acquisition. Whereas 23andMe was essentially an entertainment company and was regulated as such, which is what enabled that unfortunate situation to occur.

> But then at the end it added a "Fun fact" that unicode actually does have a seahorse emoji, and proceeded to melt down in the usual way.

To be fair, most developers I’ve worked with will have a meltdown if I try to start a conversation about Unicode.

E.g. if during a job interview the interviewer asks you to check if a string is a palindrome, try explaining why that isn’t technically possible in Python (at least during an interview) without using a third-party library.