[pgt]

The Law of Better is Worse:

User-friendly tools prevent skilled middlemen from monetizing their expertise, which stifles adoption of that tool. So on-sellable tools that are too easy-to-use, don't get on-sold.

Some examples by contradiction: tax returns, AWS Dashboard, many programming languages.

[dammitfoo]

> Some examples by contradiction: tax returns, AWS Dashboard, many programming languages.

By programming languages, do you mean Rust?

[pgt]

Absolutely Rust and JavaScript. Rust would be a far better language if it used S-expressions and didn't try to reinvent a macro syntax. But then it would not be as popular.

In this way Lisp suffers from having no syntax, although it's a slightly different argument. When you can't have flamewars about a language's syntax, fewer articles are written about it. So instead, people will argue about the encoding of the AST - the parentheses.

Similarly, well-designed languages like Clojure, Haskell and Erlang have fewer questions on StackOverflow and older GitHub issues, so there are fewer flamewars about them (although monads are Haskell's saving grace here).

The NPM crowd are quick to ask, "Is this project abandoned?" when it hasn't had any activity for a year. In Clojure country, we dislike using libraries that haven't been stable for at least five years. As Alan Kay put it, Computer Science is very much a pop culture.

The phenomenon needs a good name, though. Perhaps the Moving Target Paradox, since developers are more likely to run after a moving target.

[bergoid]

> The phenomenon needs a good name, though.

Jamie Zawinski calls this the CADT model: "Cascade of Attention-Deficit Teenagers".

https://www.jwz.org/doc/cadt.html

[acdha]

> well-designed languages like Clojure, Haskell and Erlang have fewer questions on StackOverflow and older GitHub issues

Can you share the methodology you used to validate that this explanation is correct, ruling out the orders of magnitude larger and wider audiences which the languages with more questions have?

[the-dude]

How about SQL

[pgt]

SQL gave business analysts a way to express relational set algebra. If SQL had launched with an (arguably superior) datalog syntax, I wonder if it would have been as popular. Probably not, so it does feel like a good example. If SQL was composable (i.e. not a concatenated string), it would employ far fewer API gluers.

[jonahx]

See also: Kubernetes.

This is pretty brilliant. Did you just make this up or it a real thing?

[pgt]

I made it up: https://twitter.com/petrustheron/status/1168438193752031233?...