Science is description and explanation on top of empiricism. It is the first means by which people understand things, not the last, as formal methods came way late.
This drive to properly name things also gets into the somewhat similar debate of whether math is discovered or invented. And somewhere someone is trying to determine whether it's appropriate to call math as science, art, or engineering.
I don't think that's at odds with what the parent comment said; the reason we use empiricism and description for analyzing reality is because we didn't create it and we don't know the rules beforehand. When designing a software system, you _choose_ the rules of which things interact and which things don't, and how those interactions occur; there's no need for empiricism in order to discover these interactions. We don't necessarily need to use science to understand our software systems because we rule over them by fiat and can choose to design them in ways that make it easier for us to understand them.
The problem he was proposing we solve is computing with heterogenous “machines”. This doesn’t preclude the regimented organization you are favoring, above.
Please see my other comment on call-by-meaning.
We already have reinforcement and other types of machine learning?
Something being hard is not an argument for or against anything.
Alan Kay is misguided because he prefers a hard thing and you prefer a simpler thing?
A paradigm being hard to get things done in is definitely an argument against the value of that paradigm.
That Alan Kay guy sure is a bit of a dummy huh.
I also somewhat contest that "interesting behavior comes from the emergent runtime interaction of tons of tiny components". There can be very tight, hierarchical structures to programs designed in the way Alan Kay talks about. He is promoting clear, manageable interactions, not emergent unpredictability, which is something I am sure you came across (we all did), but I would not go so far as describe the whole model of Kay as "fundamentally misguided". He talks of "clear fences", which can be understood to refer to things like "actor-based" models, with controlled, clear messages, as done in languages such as Erlang.
Remember that Alan Kay, his team and his colleagues had a lot to do with this thing you are using to rant here :)
The internet has certainly grown and does certain things reliably, but it's also extremely difficult to control or change. For most business purposes, producing something like the internet would be a failure.
> Remember that Alan Kay, his team and his colleagues had a lot to do with this thing you are using to rant here :)
Given how difficult his ideas apparently are to convey and how little the programming paradigms that actually get used to make things embody them, I find that pretty dubious. E.g. didn't he explicitly disavow C++?
It's like asking Bezos to care about the Internet as a technical concept ("Internet? Schminternet? I don't care as long as it delivers the best customer experience!").
Kay was and still is a dedicated researcher - and he has had great influence on people like Jobs/Bezos as well - behind the scenes. There are very few parts of modern technology untouched by his ideas. He's a Turing Award winner for a reason, you know.
Re C++ - Kay's job as a researcher is not to defend whatever sh*t humans have figured out & are content with but to paint a picture of the future, build teams around it and demonstrate prototypes. He has done that exceedingly well I'd say.
I don't think either PG or most people on this forum are fit enough to judge what Kay is saying is of value or not. And I say this with utmost humility. These are scholarly researchers, who operate at the edge of human knowledge/insight. Who am I to judge Newton or someone of that calibre - question why he's interested in Alchemy? Or Kay says particular things which seem far-fetched to you or me.
Continuing with the Newton analogy - I think one aspect with CS/Software is that it is a new field - you should not apply the definition of a "Charlatan" you'd use in a more established field such as Physics. Kay's agenda is to advance a nascent field. Just like Newton was dabbling with Alchemy before the establishment of the modern field of Chemistry.
Newton earned the respect we give him with his theories of gravity and calculus. He made concrete explanations that other people were able to understand and build on - even his enemies, who were many, could not deny the correctness of his results. And the whole edifice of science, engineering, and industry is built on that work - even today, Newton's laws of motion are something people have to learn and understand, and a lot of people do understand, and teachers distinguish between people who understood them and people who did not, and test whether people's understanding of them is correct or not. People judge Newton's work all the time, and they are right to do so.
Did he spend a lot of his life on bullshit? Yes, probably. So does PG, so do any number of Nobel Prize winners. But we don't remember his name because of his work on alchemy and biblical prophecies. There's a huge difference between someone who did some productive work and some bullshit, and someone who only or mostly did the bullshit. And if you take the position that you aren't fit to judge who's a genius and who's a charlatan and you're not even going to try, you're going to get taken for a ride.
In my book Kay makes lots of sense most of the times, if you put in the effort. Maybe you see it totally differntly.
Another thing - with Newton for ex - I think the way he studied these other subjects - was still very honest, very sincere, and he made great efforts to get things right. So I'd say Newton has maintained methodological integrity throughout regardless of results (quality of efforts > quality of results). My view with Kay would be the same - I think he has high levels of integrity. I can cite why I think so, but then this thread is taking too long already.
I don't think it's a percentage, I think we evaluate people on the non-bullshit they've done rather than the bullshit. And as far as I can see Kay hasn't done much that's valuable - I mean, I think there's merit in Smalltalk, but the parts of its design that I think are good are disjoint from the parts that Kay talks about. To the extent that the things he's said convey meaning they tend to be wrong - object orientation has failed in multiple incarnations, late binding has failed, live systems have failed, etc..
> In my book Kay makes lots of sense most of the times, if you put in the effort. Maybe you see it totally differntly.
I do. Some of the gnomic statements he's made have been retrospectively interpreted to mean things that make sense and are useful (e.g. "oh, obviously he meant actors"). But as far as I can see no-one ever managed to interpret them in a way that made sense and contributed to building something useful ahead of time - it's more of a Nostradamus situation than him having actual insight.
I mean, I assume he hasn't achieved literally nothing his whole life, that at some point he's done research that contributed to something useful. But I've reached the view that all the stuff he's famous for, all the stuff that people quote, is bullshit.
> I think the way he studied these other subjects - was still very honest, very sincere, and he made great efforts to get things right. So I'd say Newton has maintained methodological integrity throughout regardless of results (quality of efforts > quality of results).
The thing is, it's much harder to judge efforts than results, so it's easy for a charlatan to look like they were making high-quality efforts. I'm willing to trust that Newton had methodological integrity because he was able to produce great results, and so I'm willing to accept that the efforts that lead to that carried over to other parts of his life (not that I think it actually matters either way - if what you're studying is fundamentally rotten from the start then an investigation with higher methodological quality is a castle on sand). You have to be a lot more sceptical if you don't have that proof that the person is at least capable of high-quality efforts.
What's your definition of failed?
Hence its success.
It is controllable and changeable exactly in the manners that are appropriate by those who should be able to control it and change it.
The internet is as distributed and decentralised as it is because it was made by a distributed and decentralized entity (thousands of companies and individuals).
Had most of them been willing to coordinate and cooperate with each other we would have a very different internet.
Erlang appeared 15 years later, and was / is quite successful in certain niches, more recently also as Elixir.
But aren't we? The modern ML revolution is programming via natural selection. Emergent behaviour from complex interactions of simple components trained by selection.
And of course, experience underlines how hard these programmes are to reason about.
However, seems to me, the key trick of these systems is that very complexity that makes them so hard to understand is what gives them their computational power.
Physicians.
"but until we start programming by natural selection I wouldn't describe it as a good paradigm for organizing behavior."
Machine learning, alternatively startups selected by market forces.
I much prefer my programs to have a sort of top down structure, more like a military instead of an economy. Obviously late-binding and dynamic behavior is many times necessary, but I would not lean in to it, in the same way I would not say, make all my variables global just because sometimes you need it.