Happens in Canada too. Calling them out can be dangerous, people have been injured.

Big "Kill them with kindness" energy.

I have my own harness I wrap Claude CLI in, I wonder if I'm breaking the rules...

If you're not paying full-fat API prices, then probably.

From what I've heard, the metrics used by Anthropic to detect unauthorized clients is pretty easy to sidestep if you look at the existing solutions out there. Better than getting your account banned.

No, they specifically said it’s only if you’re trying to build a whole other product for public consumption on top of it

If you’re just essentially calling claude -p you’re fine

Neither was either of mine, but I don't advertise them and specifically don't post them on social media

I think that's a cool project, though I found the results to be less relevant than Google.

Whether the results are less relevant or not depends massively on what you searched and whether the best results even exist in the Marginalia search index or not.

If Google is ranking small web results better than Marginalia, that’s actionable.

If the best result isn’t in the index and it should be, that’s actionable.

Well to be fair, Marginalia is also developed by 1 guy (me), and Google has like 10K people and infinite compute they can throw at the problem. There has been definite improvements, and will be more improvements still, but Google's still got hands.

Hey Marginalia, cheers. Imo fewer hands can also be an advantage.

There are no PMs breathing down your neck to inject more ads in the search results, you don’t depend on any broken internal bespoke tools that you can’t fix yourself, and you don’t need anybody’s permission to deploy a new ranking strategy if you want to.

I've used Marginalia to search for technical documentation before, unironically. Whatever it does find is pretty much guaranteed to be non-slop.

I get it, but it makes me sad.

At least I already have it in my library so, looks like I still get updates.

This is not what is advertised, and if it was there would be no money in it.

I've been alive for several decades and almost nothing I've gotten was exactly as advertised.

I'm not saying that's good, but it's consistent with reality.

Consistent with what you think of as reality. For all you know, you're being Truman Show'ed.

How do they know? Not everybody includes to coauthored by Claude. I certainly don't.

It has a lot. I find by challenging it often, getting it to explain it's assumptions, it's usually guessing.

This can be overcome by continuously asking it to justify everything, but even then...

Trust shouldn't be inherent in our adoption of these models.

However, constant skepticism is an interesting habit to develop.

I agree, continually asking it to justify may seem tiresome, especially if there's a deadline. Though with less pressure, "slow is smooth...".

Just this evening, a model gave an example of 2 different things with a supposed syntax difference, with no discernible syntax difference to my eyes.

While prompting for a 'sanity check', the model relented: "oops, my bad; i copied the same line twice". smh

I don't find it tiresome at all. What I was getting at was, even with constant justifications you need to remain vigilant.

It's almost like an emergent feature of a tool that's literally built on best guesses is...guesswork. Not what you want out of a tool that's supposed to be replacing professionals!

Interesting perspective.

I guess I'm more interested in understanding what it can and can't do.

Phone is probably the best tool for most minor online banking actions.

Not all.