I used to complain (lightheartedly) about Claude's constant "You're absolutely right!" statements, yet oddly found myself missing them when using Codex. Claude is completely over-the-top and silly, and I don't actually care whether or not it thinks I'm right. Working with Codex feels so dry in comparison.
To quote Oliver Babish, "In my entire life, I've never found anything charming." Yet I miss Claude's excessive attempts to try.
And that's exactly the point, it increases engagement and stickiness, which they found through testing. They're trying to make the most addictive tool and that constant praise fulfills that goal, even as many of us say it's annoying and over-the-top.
My own experience is that it gets too annoying to keep adding "stop the engagement-driving behavior" to the prompt, so it creeps in and I just try to ignore it. But even though I know it's happening, I still get a little blip of emotion when I see the "great question!" come through as the first two words of the response.
> And that's exactly the point, it increases engagement and stickiness, which they found through testing. They're trying to make the most addictive tool
Is this actually true? Would appreciate further reading on this if you have it.
I think this is an emergent property of the RLHF process, not a social media-style engagement optimization campaign. I don't think there is an incentive for LLM creators to optimize for engagement; there aren't ads (yet), inference is not free, and maximizing time spent querying ChatGPT doesn't really do much for OpenAI's bottom line.
They still want people to stick around and 'bond' for lack of a better term with their particular style of chat bot. Like so many venture funded money pits of old the cash burn now is about customer acquisition while they develop and improve their tech. They're all racing toward a cliff hoping to either make the jump to the stratosphere and start turning massive profits or to fall off and splat on the rocks of bankruptcy. If they don't get the engagement loop right now they won't have the customers if the tech and use case catch up with the hype and you can only tweak these models so much after they're created so they have to refine the engagement hooks now along side the core tech.
I am currently working on an agent thingy and one of its major features (and one of the main reasons I decided to take on this project), was to give the LLM better personality prompting. LLMs sound repetitive and sycophantic. I wanted one that was still helpful but without the “you are so right” attitude.
While doing some testing I asked it to tell me a joke. Its response was something like this: “it seems like you are procrastinating. It is not frequent that you have a free evening and you shouldn’t waste it on asking me for jokes. Go spend time with [partner] and [child].” (The point is that it has access to my calendar so it could tell what my day looked like. And yes I did spend time with them).
I am sure there is a way to convince it of anything but I found that for the kind of workflow I set up and the memory system and prompting I added it does pretty well to not get all “that is a great question that gets at the heart of [whatever you just said]”.
Claude at times feels like it's mildly manic and has ADHD... I absolutely prefers that to Codex...
Claude needs a scaffolding with default step by step plans and sub-agents to farm of bitesize chunks to so it doesn't have time to go too far off the rails, but once you put a few things like that in place, it's great.
You're only seeing those ads because the ad algorithm knows you. My family aren't getting Claude ads. They wouldn't know the first thing about it even if it were explained to them.
This isn't a wild theory or a novel one. It's well-established that endogenous retroviruses alter DNA and are inherited. In addition to the primary genome being modified this way, all mitochondria are symbiotic organisms inside plant and animal cells, with their own DNA, and are vital to life. Same thing for chloroplasts in plants. And then there are gut bacteria, which are vital to life, symbiotic, and directly influence evolution and the genome.
It's unclear to me why running Docker directly in Proxmox (it's just Debian) and using it like any other Docker host is a bad idea, and why this extra layer of abstractions is preferable.
Docker has security issues if you're not careful, and it's frankly kind of a shitshow out of the box with defaults. Maybe that's part of the reason. But I struggle to see how a bespoke solution like this is the right answer.
Proxmox is a hypervisor OS, and its value comes from its virtualization and container-management features. These features include being able to pause, resume, snapshot, backup/restore from snapshot, and live-migrate VMs or LXCs to another server in just a couple hundred milliseconds of downtime. Once you run docker on the hypervisor itself, you lose these features, which defeats the purpose of running Proxmox in the first place.
There's also the security angle. Containers managed by Proxmox are strongly isolated from the host, but containers running on Docker sidestep this isolation model. Docker is not insecure by design, but it greatly increases the attack surface. If the hypervisor gets compromised, the entire cluster of servers will also get compromised. In general, as little software as possible should be installed on the host.
What's the reason Proxmox can't just implement Kubernetes on the host for running docker across a set of Proxmox nodes though. I mean they implemented a system like Ceph for distributed storage.
You have a bunch of tooling that deals with apples. You have a clear conceptual picture of what an apple is and what it does.
Then someone brings you a pear. It's kind of like an apple but not exactly. Their pear however works well with some other toolscape that's beyond the shire. You want to do things with their pears.
You invent a way to put a pear inside an apple (docker in VM). That works but you lose some functionality and break some stuff in the conversion, plus now you don't have the clean conceptual integrity of your apple-only system.
Largely management, observability, and then the way that docker mucks with firewalls. Running them this way will allow proxmox to handle all that in the same way {I assume) as the LXC and VMS so automation, and all the rest can be consistent
What about running Docker inside a Proxmox LXC container? Is that a common practice? Intuitively doing that would have a lower baseline resource usage than using a full-blown VM for Docker containers.
I've been running Docker natively on the host since Proxmox 7. The only major problem was an iptables rule that I had to add so that the containers are accessible from outside. Besides that, it runs smoothly.
They're pretty great pets. We had one for a while when I was a kid. Its mom got run over and we nursed it and raised it for a few months. Instinctively used the same litter box as the cats. Hung out on the couch sitting on my shoulder watching TV. Friendly and playful. Would follow people around and play with toys.
The biggest challenge is that they basically have hands. He would climb up the kitchen cabinets, grab a box of cereal, open it up and sit there eating out of it like a toddler.
We only had him for a few months before reintroducing him to the woods behind the house. I've wanted a pet raccoon again ever since.
A former girlfriend of mine had a picture of her mother holding a Raccoon. I asked her mother about it and she said that they lived out in the woods in Minnesota and they found it on the porch when it was a baby. The mother had died or something so they kinda raised it. It was free roaming in/out of the house but they could hold it and it would also get into their food. She mentioned one time it ate a bunch of mixed nuts...but didn't like one type so it left all those in the bowl. Another time it ate an entire pie...but left her one piece ("so she wouldn't get angry").
She did say it was never really a "pet"...more like a wild animal that sometimes acted like one.
This would have been in the late 70s early 80s by my guess on her age in the pictures.
They have a lot in common with housecats, except that they are more clever. Decades ago we heard a crunch crunch sound from the rear mudroom. We looked and saw a raccoon reaching in and eating dry cat out of a box with the cat looking on enviously.
Camping I heard a crunching sound, looked out from the tent to see a racoon helping itself to granola in the back of the car. Lock your doors.
I remember reading somewhere once that baby raccoons are actually quite cuddly and tame; but that when they go adolescence, they have a hormone shift that makes them aggressive enough to be unsuitable as a pet. In the story a woman who had raised a baby raccoon was attacked by it after it grew to a certain age.
Judging by the murderous sounds you hear all night here in the summer, I would not want to be cornered in a dark alley by a gang of adolescent raccoons.
>Judging by the murderous sounds you hear all night here in the summer, I would not want to be cornered in a dark alley by a gang of adolescent raccoons.
Well if you ask me adolescent raccoons are a big problem in many of our cities, I'd be worried about such a case myself.
I've heard the same thing from my mother, whose uncle had a baby raccoon as a pet. Once he got older he became mean and would yank on her hair for no reason.
There's a Japanese anime from the '70s called something like "rascal the racoon", based on an American book, which tells the story of a kid with a pet raccoon.
I've wanted a pet raccoon since I saw this on TV in the '80s, and raccoons aren't even a thing in Europe :(
I learned about this 2 years ago when a racoon showed up in Tokyo close to where I lived. [1]
wild to think they spread even all the way to Japan because of anime. and probably south Korea now. They banned raccoons in Japan but it seems to not have caught up in SK and there are a lot of racoon pet videos from SK on YouTube)
A generation of us grew up deeply coveting a pet coon, and have never given up on that dream, really...
The shocker for me was the bit about Rascal learning not to wash his sugarcubes before eating (or actually, to rinse once, because he was OCD about washing his food). Not that itself, although fascinating and charming; the idea that sugar was rationed to the author's family was mindblowing to juvenile me.
Racoons are invading the north east of France:
https://www.youtube.com/watch?v=I2ijwZROb6g ; they are exotic invasive species: a female get 10 babies per year, and there is no predator
Then identify the "morph", the gene cluster, that did the trick in the initial species, to speed up the uplift for the rest of them.
But given how intelligent they are, and how much they learn in their short lifetimes, completely self-directed and self-learning highly curious autodidact geniuses from the moment they escape their tiny egg, a race of long lived octopuses without learning brakes might be an existential threat for us.
Or they might create a magical ocean civilization. There are octopus species everywhere. Deepest ocean floor, under the polar cap, even a "land" species that lives on the beach. Everywhere from sea level down.
Death follows mating for the males and the eggs hatching for the females. Not an uncommon behavior, all sorts of plants (ie agave) and animals (ie salmon) have a similar strategy.
Meaning the strategy works very well, and would likely be controlled by very different genes, depending on the species.
Changing this "death gene" via a breeding program is likely to fail, but could perhaps be accomplished via gene editing like CRISPR.
Another approach would be delaying sexual maturity. Doubling the lifespan to something like 4-5 years in the common octopus would have very interesting results.
However, doubling the lifespan would effectively halve the number of offspring, meaning these mutants would be less competitive, compared to their wild relatives.
This could be seen as the needed safeguard to prevent a race of highly intelligent mollusks from taking our planetary domination crown.
So all in all, delaying sexual maturity seems to be the better way to go for your project.
Yes, delayed reproduction would be the first step.
And simultaneously for post breeding longevity (however short that starts). Which can be done by post-selecting the young based on the phenotype of the parents. (No need for any harsh selection, simply choosing which young remain in the breeding pool, vs. which go there own way.)
Their short breeding cycles are a boon, since that allows for faster generations, and even small lifespan improvements would reflect significant change.
The high number of species is also a tremendous advantage.
All done at scale. Millions of octopus, across dozens of species, to efficiently select from as much existing genetic diversity as possible. Not just for faster gene clustering, but to gain different insights from different species that can be transferred, via CRISPR. As you noted.
And finally, also selecting for individual intelligence and social collaboration. They are unique in being an extremely socially intelligent animal, with high cross species cognitive understanding. There preference for solitude does not reflect any lack of social awareness. So this is one of the easier and potentially rewarding challenges.
Since effective intelligence in practice is a (literal) product of individuals and collaboration. There is tremendous opportunity for gain of function.
—-
Of course, I will need a vast underground laboratory under a tropical island for all this. A not-too-dormant volcano for cheap geothermal power. And a comfortably furnished submarine of my own design, for research forays, with a streamlined exterior inspired by the profile of a jetting octopus. Christened with an ominous name.
As for other resources? Well the ocean has infinite untapped resources, and I will soon have infinite assistants.
Ok, some of the latter might be me getting ahead of myself.
Such records shouldn't exist in the first place. I agree they shouldn't be exempt if they do, but let's not just accept that it's okay to have a fleet of cameras recording us 24/7 everywhere we go, managed by a private entity, accessed freely and without any probable cause by local and federal agencies who don't even communicate with each other.
It was argued that making the Flock data public violated everyone's privacy. It's important to stress and remind everyone the privacy violation occurred the moment pictures are indiscriminately taken, processed by AI, and stored for every single car that passes by. Not to mention family homes, pedestrians, and other things being captured in the process.
We are only a couple steps away from doing the same thing for pedestrians. Why not just take pictures of every single person walking by now? This already happens in some places. Flock is paving the way to make it a government sanctioned mass surveillance program.
I would have done the same thing. I know how to build software in a dozen or more languages. I've done it manually, from scratch, in all of them. I don't know Rust. I have no immediate plan to learn Rust. I vaguely know that Cargo is something in the Rust toolbox. I don't have it installed. I don't particularly want to learn anything about it. It's a whole lot easier for me to tell the LLM to figure that out.
I might learn Rust some day. At the moment, I don't need the mental clutter.
Well, fyi because it is really simple: if you have rust installed, you have cargo installed too. And to run a project you type “cargo run” from the base directory. That is all.
I don't doubt that the process is ultimately fairly easy, but it always looks easier from inside the bubble. You have to know that Cargo is part of Rust; what the best way to install it is on your system; what tooling you should use if you don't want headaches later; what to do if you don't want to or can't install system-level packages; what minimum versions of Rust and Cargo are assumed by application developers this week; what additional dependencies are required; etc.
It's my understanding that building Rust applications still requires a C toolchain, and packages are still going to be dependent on things like having the openssl dev headers/libraries installed. That's fine, that's normal for building software, but it's not as trivial as "just git-clone this Rust source repo and run one command and everything will work".
I'm certain I could get up and running quickly. I'm also certain I'd have to install a bunch of stuff and iterate past multiple roadblocks before I was actually able to build a Rust application. And finally I'm certain I could get Claude or Codex to do it all a lot faster than if I muddled through it myself for half an hour.
Yes. The term has been around for at least a few decades, but only became somewhat more widely used in the past decade. Only really known by people who spend a lot of time on the command line.
To quote Oliver Babish, "In my entire life, I've never found anything charming." Yet I miss Claude's excessive attempts to try.
reply