Colin Roberts

@_JaidenWoods @ThePrimeagen Have I ever once said that I believe people should, themselves, write 0 lines of code? No. You’re actively trying to make an argument where there isn’t one. Why in the world would I propose doing something in an objectively worse way?

@Autoparallel @ThePrimeagen For example right, say you have an algorithm where you know the data structure is going to have issues with cache locality so the architecture needs to be optimised for that. There is no way for you to do that with llms, you can argue with it and go back and forth.

This actually makes me sad. Programming is such a fun thing to do in of itself. The completion of the thing task is the ultimate goal, but the road to get there is enjoyable! Even as i investigate loops and think deeply about how to deploy AI in a real manner that produces real and good code I still make time for 45 minutes to 3 hours a day trad programming. But regardless of AI being perfect or not, being able to produce quality work or not is not even part of the argument. I just like programming and I think that loving the craft of building the thing itself will ultimately make the thing you are building better. Because you care deeper about it. Its not about a pretty dress & lipstick you put on it, but the deep thought care you put in to the things no one sees or perhaps its so good they just don't notice. I think that is where the best software is made.

@AstraKernel One can only hope 😂

@Autoparallel I am not sure which drama😉 but the AI got lot better than 3 years before. But wait, soon social media will find different topic to upset about :)

Wow, the latest AI programming drama is really just the same drama there has been for the last 3 years. Can we find something else to be upset about so my timeline can be less boring?

@_JaidenWoods @ThePrimeagen I made this comment elsewhere, but I’d be curious if you think using a lathe is still woodworking? I mean, a machine is doing basically all the work for you. Where is the line drawn?

I’m not really here to start an argument, but your point and analogy don’t really count to me if it’s just going to some pure semantic driven strawman. Programming involves a lot of different elements. Some of those involve the code being written, compiled/interpreted, and ran. If I sit down and design an algorithm in pseudocode, am I programming? I’d think so, even though it wasn’t compiled or ran. It’s still designing a set of instructions to be ran by a machine.

@lauriewired Do people not like woodworking if they moved from a chisel to a lathe?

I’m convinced that a large % of programmers don’t actually like computers. As a side effect, are also perfectly happy to throw away their reasoning to a model as soon as they can. I don’t get it, at ALL. Don’t you *LIKE* understanding the magic of the machine? You do realize hand-programming (I hate that I even have to specify hand now) is fun…right?

@ZeroDarkDev What can’t SQL do for you?

I keep reinventing the wheel doing data analysis for my research. Then I remember a relational database can probably do 95% of what I need with SQL alone. Maybe I've been overthinking this?

@Qivshi1 Arxiv has Tex for all papers.

its so fucking annoying the only way to get academic articles into llms is pdfs which they suck ass at parsing

@bryan_johnson Probably focusing on friends, family, and relationships. You do a good amount of that already, but that’s probably going to be the best for you.

Eager to hear your life advice. What should I be doing?

@ThePrimeagen It’s fun because it’s neither of these things.

just found this gem

@dcbuilder This is an interesting way to battle the spoofing problem. How “near impossible” is it quantitatively?

Octet uses the TEE in your phone to assert liveness of the device, but also uses many of the phone's sensors to prove where you are, GPS coordinates are easily spoofable, but if you couple that with gyroscope, accelerometer, magnetometer and other sensors you get an input that's near impossible to spoof coherently over time.

There is proof of human, there is proof of finance, zk kyc, zk email, zk tls, but an important primitive was missing: proof of location. Very happy to announce that I invested in Octet as an angel and I'm also an active advisor to the team.

@luhelminger My bad, it's Telegram that doesn't have it by default. You have to choose secret chats. WhatsApp does this better for sure. iMessage also does this by default, and Signal too.

@Autoparallel WhatsApp is encrypted by default. Not sure if you can even turn it off.

Why is Apple spending money on this ad when users don’t care about privacy 🤔

Okay. Bit confused as to the aside explaining privacy to me when it is clear my intent previously was shorthand only to then say “Claude in a private way …” Anyway, I’d genuinely be curious to see how many people would choose Claude-but-private over just Claude even with no downsides. Again, people choose to not enable encryption on messaging apps even though they lose nothing by doing so and gain some amount of privacy or security.

There is no such thing as a privacy or non-privacy app for two reasons - Privacy is a feature not a product or app - Privacy is a spectrum. Nearly every app has some privacy. If you could offer the functionality of Claude in a private way without noticable downsides you would have an unfair advantage.

I just look at the adoption of non-privacy apps and a lack of adoption comparatively for privacy-preserving applications. A small fraction of potential users avoid LLMs because of privacy concerns. I think given the option of a desirable application, people would take a private version over the non-private, but even still this isn’t totally true (see, i.e., usage of Signal compared to basic SMS/RCS or even how few people enable encryption on WhatsApp).

@Autoparallel Where do you look for evidence? Any data source you keep track?

They have tried all of those things over the last decade. It’s definitely an attempt to see if this can stick. Maybe it will work. I have no idea. I also haven’t seen evidence of a large percentage of the population craving for privacy and, if anything, using tools that remove more and more privacy constantly.

@Autoparallel That’s not the point. They could have gone with anything: AI, design, etc But they decided users will care if we say Safari is private. It doesn’t matter if you think Safari is, it shows Apple thinks they can get back users with privacy.

Every time I am thinking of dealing with git rerere i know I’m in for pain.

Ultrasound tomography has been a thing for a long time; this isn’t really a “new imagine machine”. It is still fascinating and amazing that they’re pivoting to build this. Would be fantastic to have drastically cheaper full-body imaging. It won’t ever compare with MRI unfortunately.

🤯 Midjourney -- yes, the AI image company -- just shipped a brand new type of imaging machine. 🤯 - 100x faster than an MRI. - 10x cheaper. Full body scanned in 60 seconds instead of an hour in a tube. Ultrasound based, MRI-level resolution. And it's real -- not a concept, a working machine. You step into a shallow pool of warm water, a ring of half a million sensors sends sound through your body from every angle, and ~60 seconds later you have a 3D map of your insides down to a fraction of a millimeter. No radiation, no tube, no lying still. They're not even building it as a hospital machine -- they're building a spa. The scan is a side-effect of a place you'd want to hang out anyway. Lastly, it is built by 9 people. NINE PEOPLE. You can just do things.

@homieomorphism It stopped?