Yes, absolutely! I definitely want to look into this, although it's not the top of the current roadmap.
To me, the first step is going to be to really work through and trying to get this right. Do user studies. Watch people write code in this. Watch people with lots of experience, and people with none get tossed into a project written in the LOGOS and told nothing.
Once the language surface is more solid and not as likely to go through major changes, I want to focus our efforts in that direction.
Don't take this the wrong way, but my understanding was that you're vibe coding it?
If that's the case I'd do this from day 1, your parser should be a 1 to 1 mapping of some text to code, this you can easily and rigourously test, then if you want to, you can do other stuff on top
I personally think this is an uncharitable reading, you can have a different internal benchmark or standard you want for yourself vs others
From a purely consistency perspective I don't think you're incorrect, but humans aren't purely consistent
We are able to accept that our personal preferences aren't the same as others and still like, respect or love them anyway
I read the GP as stating:
- he wanted to work for a place that made him happy
- he voiced that pleasure to others, "I'm glad I work at a place I find inspiring"
- they took that as an implicit attack on them
There are at least two parties to a conversation, each of them gets their own opportunity to interpret what occurs
It sounds like in this instance they interpreted his position much more negatively than he intended
Now to answer why is in my opinion is much more complicated and I honestly wouldn't hazard a guess without either being there or knowing both parties very well
Just FYI. You're right, and I probably could have phrased it better, but I wasn't talking about this post.
Most posts are "I worked at a company that did stuff I really liked, and was honored to work with some really inspiring people."
That's usually enough to cause people to assume that I'm insulting them.
I do my best to not be offensive, but some folks live in a world, where everything is a personal slight, and there's really nothing I can do about that.
I'm curious, I'm working on a multiplayer game and I'm wondering if this would be a cool new alternative platform for it to target, would you be open to a chat?
This is a really interesting observation, as someone who feels disquiet as the obsequiousness, but have been getting used to just mentally skipping over the first paragraph that's put an interesting spin on my behaviour
I'm genuinely wondering if your parent comment is correct and the only reason we don't see the behaviour you describe, IE, learning and growth is because of how we do context windows, they're functionally equivalent to someone who has short term memory loss, think Drew Barrymore's character or one of the people in that facility she ends up in in the film 50 first dates.
Their internal state moves them to a place where they "really intend" to help or change their behaviour, a lot of what I see is really consistent with that, and then they just, forget.
I think it's a fundamental limitation of how context works. Inputting information as context is only ever context; the LLM isn't going to "learn" any meaningful lesson from it.
You can only put information in context; it struggles learning lessons/wisdom
Not only, but also. The L in ML is very slow. (By example count required, not wall-clock).
On in-use learning, they act like the failure mode of "we have outsourced to a consultant that gives us a completely different fresh graduate for every ticket, of course they didn't learn what the last one you talked to learned".
Within any given task, the AI have anthropomorphised themselves because they're copying humans' outputs. That the models model the outputs with only a best-guess as to the interior system that generates those outputs, is going to make it useful, but not perfect, to also anthropomorphise the models.
The question is, how "not perfect" exactly? Is it going to be like early Diffusion image generators with the psychological equivalent of obvious Cronenberg bodies? Or the current ones where you have to hunt for clues and miss it on a quick glance?
I'm 100% with Tonsky on this one, it's something that's frustrated me for years and I think it fundamentally boils down to a lack of respect for your user.
I'm going to be careful here, this doesn't mean that there aren't times when we can't avoid bothering our users, but I think we resort to doing so far too often, we don't apply the axiom of "if this person was someone who I deeply respected and would feel real bad if the contacted me and asked me to justify bothering them, would I still do this" as our actual test.
I'm not saying this is uniquely our decision, this entire process might be out of our hands, but in my opinion it bares thinking about and weighing up appropriately.
It's very easy to say it's only 5 seconds, if your software serves millions, that's a lot of people and even if you don't, a couple of 5 seconds here or there adds up very quickly.
I'm not here to berate or point fingers, but we also are users of each others software, so I hope that at least on that level we'll try to do better :)
I agree with the sentiment of the post, but sometimes it would be too ‘expensive’ to implement a solution to some of these problems.
E.g. on user accounts: nobody likes them, but when Uncle Fizz asks customer support for his data because he lost his phone again, a synced user account would be the simplest way to help him.
Updates should either be automatic and transparent, or it's indeed on you to keep track and decide whether to update. I do agree that NPM packages are freaking annoying, every package now needs to tell you something when you install or update.
Same with What's New modals, some people will benefit from learning these things (notably power users?), but they'll annoy others.
Notification dots are idiotic.
So… striking a balance where we can? Otherwise most users would be left behind, as if we'd given them a terminal and said APK installs and updates things.
I'm not sure what the solution to all this is, but I like Tonsky a lot, and it's a great blog post.
> Same with What's New modals, some people will benefit from learning these things (notably power users?), but they'll annoy others.
I think power users are most annoyed by those modals. It prevents them from doing the exact thing they were planning to do. Instead, they'll have to reinterpret what the application is telling them, consider it to be irrelevant (most of the times), and then pick up whatever they were planning to do. This creates friction.
I don't need the application to tell me a sidebar was introduced. I see that immediately because it differs from the layout I'm already used to. And then I'm annoyed they added the sidebar, because it takes up space without offering relevant new functionality.
Not sure, I will publish this in different modules and add most of the content over time but first release might be around early january? really depends on my workload on my real job. I might release it earlier as proof of concept, but not planning this right now.
I've been thinking about this a lot recently, I think it's a mixture of the effects of power laws we see in larger orgs having more revenue and their biggest concern is not having any issues that risks the reputation etc they've built.
Smaller more nimble orgs that are still finding their feet have an uphill climb, but the cost of mistakes is lower
This statement reads to me to be heavily hindsight biased.
CERN wasn't exactly a place filled with idiots, yet the article even says that Tim Berners-Lee's boss thought the concept was a little eccentric and only gave in because Tim Berners-Lee fought for it.
Unless you're saying the concept is simple? In which case yes, most brilliant ideas that are hard to have are made by elegantly combining things to make a "simple" result.
The really annoying thing about those ideas is you sit there and kick yourself thinking, "that's so simple, why didn't I think of that"
There's a very different possible future where he instead went private and sold it, and I honestly have no idea how to work out how successful the web would have been in that world.
A good chunk of the web's impact is it was how easy it was to adopt, so I doubt we would have seen as much success as we do see now, as one of the bedrocks of our current ecosystem.
We might even have seen a similar situation to unix and linux, where a theoretical proprietary web that was released eventually was rewritten in an open-source format, but with lots of fragmentation of the ecosystem.
Surely hypertext and Ted Nelson's Project Xanadu were well known to most people in the field by the time Berners-Lee did his work?
If Ted Nelson hadn't been so obsessed with making it pay we might have had the web sooner. Whether that would be a good thing or not is debatable though as the Internet was not available when he started.
Hypertext systems precede the web, I was using hypertext documents on CompuServe in the late 80s. It's hard to disagree given what was available at the time that putting the hypertext documents on another kind of network was a natural progression.
If it's supposed to be a learning resource, that would allow students to process rust and at least begin to understand what it means?
reply