3 ms·
> Apps have been "building themselves" for decades: it's called a compiler. To your point, can I call the cement mixer a self-made building material that resul
by jimmy2020 5y ago
> Apps have been "building themselves" for decades: it's called a compiler.
To your point, can I call the cement mixer a self-made building material that resulted in an accumulation of 4000-year-old of knowledge?
Every known industry has different levels of production to reach the final product. Which may be a tool builder to automate the industry. Exactly the same way the building evolves to the prefabricated. The limitations are with the raw material which is clearly to the advantage of the software industry.
> Meanwhile, better abstractions, programming languages, type systems, mathematical advances, training programs, techniques, and architectures have been raising that water level every year for a long time
I totally agree with all of the above. But that can be done along with investing more in the essential no-code tools. And if some has decided to go further good luck.
The argument about we need to do more so please let's stop the entire direction of automating because somewhere they still using CRUD, old-fashioned and unmaintainable code is similar to let's stop space discovery because we still have problems on the planet earth. We can do both.
- feoren 5y agoI feel less like I'm asking to stop space discovery until we fix Earth, and more like I'm cautioning against the claims of crystal healing and homeopathy. I would have no issue with a well-conducted, good-faith research study into crystal healing -- after all, simply coating a surface in silver dramatically increases its anti-bacterial properties, so clearly some "dark-ages alchemy" level claims are actually true. But with our current priors, given a random claim about crystal healing, what is the probability it's backed by solid research vs. the probability it's quackery? Similarly with no-code. Should we stop trying to build no-code solutions? No. In fact there are limited used cases where it goes well, such as the Blender material designer. But with our current priors and given a random claim about what some no-code solution will be able to do, the probability it's BS is very, very high. Maybe that's actually the answer: the place for no-code is not in replacing general-purpose programming languages, but in small, limited-purpose DSLs like shader languages (and even then, only in specific contexts). SQL is an example of such a thing, so if someone came out and said "I have a great no-code solution that's meant to replace annoying repetitive SQL queries in a limit set of circumstances", I would take that person much more seriously.
- jimmy2020 5y agoI appreciate your ironic tone.