Hacker News new | ask | show | jobs
by asdflkj 5847 days ago
Here is the real problem with desktop Linux: nobody wants to take the only thing that Linux is good at, and make it better. I'm talking about hackability. Ubuntu and similar systems make it worse. There is too much ugly, incomprehensible and plain unedifying kludge between the neat Unix core and the whatever shiny crap is in fashion this year. If a minimal distro is a pool with no shallow end, then Ubuntu is a pool with a shallow end, and barbed wire between it and the deep end. It's hard to make a pool with a shallow end that gradually gets deeper, but as far as I know, nobody is even trying.

Not every potential user is an idiot. There are plenty of people who would learn, if the learning process felt more like lifting weights and less like having drain cleaner poured on your face.

3 comments

The real problem with desktop Linux is that people don't care. They use Windows because they have to. They use Windows because everything works with it and everything comes with it preinstalled. Everything works with Windows because people make sure that their things work on Windows because that's what everyone uses.

I don't think anyone interested in "hackability" feels like it's analogous to having "drain cleaner poured on your face". There may be ways to make cool scripting more accessible, but there's nothing particularly contrived about it IMO.

The bottom line is that people don't care that much about computers. If Windows lets them do everything they want to do, then there's no problem. And all most people want to do is use Facebook, Excel, the proprietary program needed for their profession, and email. All of these work fine on Windows and everyone knows how to get to them from the Windows interface. They know the ins and outs of Excel and for most Excel jockeys, OO Calc isn't going to cut it, because they have to relearn some shortcuts, button locations, etc. to use Calc, but they already know how to use Excel and can hit the ground running, so they prefer Excel. Same with almost all other software.

So the real reason the Linux desktop isn't going anywhere is because nobody cares. If you want Linux desktops to proliferate, you have to give people a reason to care; something new and specific that a normal person would think is cool and buy-worthy. For most people, technical arguments like "The TCP/IP stack is great!", "iptables is great!", or whatever, they don't care about that.

I think that this has to do with flaws in the UIs themselves though. Having one sellable point is a great plan, but not if the rest of the user experience is poor. People should never have to open up the command line, yet you have to do it all the time in order to get any use out of a distro even like Ubuntu.
I'm not talking about "most people". I'm talking about some significant subset of smart people who would love hacking, but don't know what it is yet because they weren't given a proper opportunity. "Most people" will follow when programming literacy becomes the new math literacy.

  Not every potential user is an idiot.
Sure. He may be very smart person—just in the different area of expertise. For him computer is just a tool. Like not everyone wants to learn about shifting gears, tuning carburetor or operating the choke just to be able to get from point A to point B by a car, not everyone wants to learn hot to tweak a computer.
Here is why this common metaphor is wrong: the only useful thing cars do is get you from point A to point B. Mucking around with the innards is pointless unless you like doing it. Hacking a computer could be both fun and useful for a great variety of smart people. Furthermore, a car is a physical thing with rigid physical constraints on its design. If you want to tweak your car, you need to do a lot of stupid busywork with your wrench before you get to anything interesting. You need to have a wrench, and a garage for that matter. With an operating system, there are basically no constraints beyond the imagination of its makers.
> If you want to tweak your car, you need to do a lot of stupid busywork with your wrench before you get to anything interesting.

Sounds plenty analogous to the many parts of the compiler or interpreter toolchains in place in modern OSes.

> Hacking a computer could be both fun and useful for a great variety of smart people.

I don't understand. Could you give concrete examples of "hacking a computer" that could be fun and useful for a great variety of smart people?

Writing up a bot program to gather tons of statistical data so you can do your job as a marketeer thousand of times faster than your peers.

Crisscrossing traffic patterns with a database of customers so you can efficiently landscape all your clients and make more money instead of waiting around being stuck in traffic.

Build and program swarm bots that clean the floors of your client faster so you can jump to the next client.

Write A/B testing programs for your restaurant chains so you can increase earning from your customers by incrementally trying out ways to get more customers, sell more foods, etc.

Programming, however, only tended to be used by programmers.

There is already a "programming" environment used by lots of non-developers: it's called Excel.
Of course, only those with a programming skillset have any chance in hell in realizing the benefit. The marketeers, the doctors, and the mac artist would care less. All they want is a shiny UI with their inefficient workflow.

I suspect Usability designers would probably only care about easy to learn interface over the hard to remember but extremely efficient keyboard commands.

Emacs is a perfect example of this. Extremely efficient and very good at what it do, but it requires a lot of learning for the user. However, the benefit will paid off for the learning curve thousand of times over.

Programming is only an impressive-sounding "skillset" and not a simple skill, like riding a bicycle (at least in its rudimentary form), because of how awful the interfaces are between simple code you write and things happening that make a difference in your life. Maybe this problem will eventually be solved not by a better Linux UI philosophy, but by things like Heroku.

"A better UI philosophy" wouldn't even have to be anything new. It's just the classic modular Unix way--the thing that makes shell scripts so easy and satisfying. It should have been extended to GUI. I would guess that someone tried, but it didn't catch on. The idea is too obvious.