Part of it (and I think this goes to what the other says about metrics and building for the masses) is a lot of the problems we were solving two or three decades ago are solved.

We used to have to hack things together because nothing worked. There was no consistency, standards were all over the map, software solutions for most things didn't exist, and running software on the major vendor ecosystems was heavily silo'd.

Dozens and dozens of technologies changed that. Web protocols and virtual machines broke siloing. Search engines and community forums made discoverability much, much easier. We passed the tipping point where hardware was only valuable if it could be connected to an ecosystem, so engineers started building standards like USB, wifi, bluetooth, and a TCP-accessible interface into everything. And an army of open-source hobbyists wrote hundreds of thousands of libraries to "X but in Y."

So hacking itself has moved away from problems like "get a telephone multiplexer to translate a bitstream to colors on an anlog TV" and towards "What nine libraries and five gadgets will you glue together to" (for example) "let your refrigerator track your shopping list," or "How can you make setting up email not feel like hacking your left arm off for the average non-computer person?" Because those are the kinds of problems that are still unsolved.

It's a different kind of hacking requiring curiosity at a different level and sometimes a different problem-solving skillset (less experimentation, more curation and cataloguing).