More Than a Million Bytes Should be Enough for Most Computer Programs
Those of us who grew up with Mosaic, a Web browser from the National Center for Supercomputing Applications (NCSA) prior to Cello, Lynx, tkWWW, and others, knew an era when the Web was akin to Gemini Protocol, sans encryption. Notice how, like Geminispace, it all started with CERN and NCSA. Like the Net, which didn't start with GAFAM but academic institutions and the military. It wasn't always about greed, bloat, ad, spying...
NCSA Mosaic was first released in 1993. It was coded by Marc Andreessen and Eric Bina. You can still find an archive of its Web site, the binaries pages for XMosaic, and the last version. We're talking about the mid-90s when those things could fit on a floppy. You could surf the whole Web with them.
Now, 30 years later, the download size for the same thing on Android is around 10 megabytes. Like other dumb things (think containers), we've added bloat while improving nothing at all.
The same can be said about text editors, images editors, and all sorts of other things. The trajectory of library-inherited bloat seems universal. There are still loads of bugs and the main effect is, we need more hardware to run or do the same thing.
Who said computing would improve over time? The improvements are subjective. █