There's just so much wrong with that response I don't hardly know where to begin...

1. Thumb drives are often not partitioned. There's no particularly good reason to do so. Just put a fat32 right on the bare drive and it works fine. Most people don't know what a partition is. In Windows, you only really need to deal with partitions when setting up a new hard drive, which most users never ever actually do because the drive comes pre-loaded from the store.

2. "Download this confusingly-named external piece of software" is not really a good counterargument to "Ubuntu has some spots that will still trip up a non-techie user".

(I know that G=Gnome (or Q=Qt), PART=Partition, ED=Editor makes perfect sense to us nerds, but it does not to normal people)

3. "Choose between either the Gnome one or the KDE one or some other partition manager" does not make the user happy. They just want to format a damn drive, they don't want to explore the ideological rift between KDE/Qt and Gnome/GTK. And to head off a potential response that they should use the one that corresponds to their desktop environment: They should not have to know what their desktop environment is.

4. Even disregarding all of that, this is just one thing. There are other little things like that. Even if each little problem has a solution, it's still a separate solution for every little problem that shouldn't bloody well be there in the first place. About 50% of the time, my laptop decides that networking is flat-out disabled when I bring it out of sleep. About 25% of the time, gnome-power-manager dies and it won't go to sleep when I hit the power button. Every once in a while pulseaudio dies and nothing will play sounds. Changing my Windows Networking workgroup requires a trip to the config file in a text editor. There's no obvious way to change the number of desktops without a trip into gconf-editor.

These are all just little things off the top of my head. There are probably more I've forgotten, and there are certainly more I haven't personally encountered but others have.

I dunno. Maybe you can. I'm too much of a pussy to try it.

And, well, the kernel they install has worked fine for me. That's the upside to it--I haven't *needed* to recompile my kernel. It's nice not to have to go through and decide each individual damn "Do you want to install driver for X?" question in make xconfig...

s/a medium sized installation of Slack machines/a single Slack machine/g

That's why I switched over to Ubuntu. Firefox kept crashing and wouldn't run Flash Player 9, presumably because some library somewhere wasn't updated enough. Every time I wanted to upgrade something, I'd spend days in Dependency Hell trying to track down all of the other software that I needed to upgrade too to make it work.

Eventually I just said fuck it.

Problem is that Slackware is so low-level that to do anything with it requires the knowledge and confidence to fiddle around with the system. With distros like Ubuntu, people who would think nothing of recompiling their kernel under Slack are uncomfortable doing something that might put the package management database into an inconsistent state.

This one just came up for me: Formatting a USB thumbdrive. Required dropping to a shell to run fdisk and mkfs.vfat.

(Which I could do, since I was a Slackware user from about '96 until this year, but I was expecting the procedure would be along the lines of 'Right click, select "Format"', not "Determine the device name, fdisk it, and run mkfs" like on Slackware)

True enough. But Uncle Max might, because he thinks he's good with computers but actually isn't.

And you do realize that you just responded to my complaint about Linux not being user friendly by telling me to download some *extra software* that has a name which--while it makes perfect sense to us geeks (G for Gnome, Part for Partition, Ed for Editor)--would confuse the hell out of a normal person, to do something that's a fairly basic part of using a computer?

That does not help your argument.

I did find gparted as a solution when searching, but for me, doing it from the CLI was easier. So I did. I'm talking about the people for whom CLIs are scary. Something like gparted is gonna be just as scary.

So have it only work on unmounted volumes, or have it only work on removable media. There's a much bigger chance of fuckup when you have to figure out it's /dev/sdi from the dmesg or df output and then type that in than when you're pointing right at the little icon that looks like a removable USB stick.

Hell, I'm a professional programmer, so presumably I could do it myself.

Go ahead and do a google search for "format usb ubuntu". You'll find some pages returned that are people posting in the appropriate forum that this would be a good feature to have. From February. I.e., it's been suggested to the developers, and apparently, like me, nobody else is motivated enough to fix it.

And really, this is the crux of my argument. Yeah, making it so anyone can fix something is, in theory, how open source improves. But in practice, most people don't care enough to work on the little things that are the key differentiator between mediocre-to-bad user experience and great user experience.

More like Linux for the geek who'd rather spend time USING his computer than fiddling around with it. Of course, saying this, I've just spent about four solid days fiddling with my Ubuntu system so things like control-U work properly like I was used to with Slackware (it means delete-line, dammit, not underline).

And yes, that's who Apple targets as well.

No it's not. A lot of it is getting there, but there are still to this day things where Linux just falls down on usability. And I say this as someone who's been using Linux (and, until this year, *Slackware* Linux) as his primary OS since the mid-90s, so it's not just that I'm dumb and can't figure things out.

Quick example: The other day, I needed do format a USB drive because there was something messed up with its filesystem and it thought it has less memory than it actually does. I'm running Ubuntu now, so I assumed that there was a nice easy Gnome way to do it.

Yeah, there's not.

I right clicked on the drive, and I thoroughly searched the Nautilus menus and there wasn't a "Format" option. Eventually I gave up and googled it...and learned that the only way to do it is apparently to drop to a shell and run fdisk and mkfs.vfat by hand.

Even if you dispute my central premise that Aunt Tilly can't handle shell commands, you've got to at least grant me that forcing Aunt Tilly to handle shell commands where a one-character typo (e.g., /dev/sda instead of /dev/sdi) could completely wipe out her hard drive is not particularly user friendly.

It's a little thing, but Linux is riddled with these little examples where the user is tooling along happily with Gnome wrapping them in a nice warm blanket of user friendliness and then they suddenly get kicked in the balls by the underlying Linux way of doing things when they have to set up slightly weird hardware or configure a program that doesn't have a nice graphical editor for its config file.

It's small stuff, yeah, but good user interface design is all about sweating the small stuff. Apple gets that. Microsoft doesn't, really, but they're close enough for most people. The open source community, for the most part, doesn't. Certainly some do, but for every coder out there who does there are ten going "So? What's so hard about fdisk and mkfs? People should learn to use the shell anyway, it's way more powerful." And they're the more prolific coders.

For years, GM, Ford, and Chrysler have been making multiple lines with basically the same car in them. E.g., the old GMC Safari van was the exact same vehicle as the Chevrolet Astro, just with different chrome slapped on the front. If you've been listening to the news lately, you may have heard that one of the plans the Big Three have to rescue themselves is to knock that shit off and simplify their car lines.

An example from my own personal experience: I was looking to replace my car and decided I wanted a hybrid. I went down to the Honda dealership to test drive the civic hybrid and the dealer started trying to sell me the non-hybrid civic since it was cheaper and he thought I'd be more likely to buy that day for a lower price. But if you took away my base "I want a hybrid", then my choice went from 'Civic vs Prius' to 'All of the small four-door cars currently on the market'.

Long story short, I bought a Prius.

Also, for a while, Apple had the Macintosh Performa, the Macintosh Centris, the Macintosh Quadra, the Macintosh Powerbook, the Macintosh LC, the Apple Workgroup Servers, the Color Classic, and probably a few others I'm forgetting, all being sold at the same time, all with a variety of different model numbers and configurations, some of which were the same machine internally but with different nameplates on the front for different markets. One of the big changes Steve Jobs made when he came back was to simplify that down to: iMac/Powermac, iBook/Powerbook (and then later iMac/Mac Pro, MacBook/MacBook Pro). Simplifying the line increased the overall Mac sales. People could just choose between the consumer and pro model, laptop or desktop. They didn't have to choose between high-end, lower-high-end, midrange, upper-low-end, low-end, in desktop, laptop, or tower configurations, or maybe the Classic line if they wanted a monitor built in, etc.

There was a great show on RadioLab about this subject the other day where it's explained a lot more eloquently than I can, so I'm going to link you to there and stop rambling.
http://blogs.wnyc.org/radiolab/2008/11/18/choice/

First off, I want to point something out: "Apple's throne" was achieved in less than two years, starting basically from zero, when competing against companies that have been in the cell phone market since the 80s. Keep that in mind when criticizing Apple's business strategies.

One SIGNIFICANT subset of the industry where open source is not the default way to develop software: Industries where the user interface matters. Think about how many times you've heard the phrase "As easy to use as Linux".

There is a lot of empirical evidence to refute this. Customers DO NOT want choice. One of the big complaints about Linux is that people have to choose between Ubuntu, Redhat, Slackware, Debian, Kubuntu, Fedora, LFS, Gentoo, etc. Or maybe FreeBSD or NetBSD. And on top of that, Gnome or KDE or something else. When faced with too many choices, the reaction amongst most humans is give up. One of the reasons Ubuntu has been so successful is that (unlike, say, Slackware) you don't have to go through and choose which programs and window manager/desktop system you want.

One of the biggest wins by far of the App Store is that there is a certain minimum quality level needed to be in it. If they opened that up, it would turn into something like SourceForge and it would be impossible to find the good stuff amongst the chaff.

But I thought you said choice was good? ;)

Honestly, I prefer Apple to have competition. Keeps 'em honest.

All of the other smartphones are already a lot more open than the iPhone, and (with the exception of Android) they've been around a lot longer. Apple's still whuppin' their asses.

Wow, you don't understand Linux people at all, do you? There is a certain sort of person who will try to install Linux on anything that stands still in front of them for too long. The only computing hardware that people won't try getting to run Linux is computing hardware that's already running Linux. And even then, they'll try to swap in a *custom* version of Linux. It's what they do. Making the iPhone more open would just have made that happen more quickly.

There's another way to look at that.

Nokia: A huge lineup of phones for sale, for any carrier, with a history of mobile phone sales going back to the 80s. They have phones targeted at every price point, every demographic, every market.

Apple: Has one cell phone model (available in two configurations), which it has been selling for a year and a half. In most countries, it's only available with one provider, and it hasn't even been *available* (legally, anyway) in most countries until the middle of this year.

And despite that, according to you, Apple's already matched 1/4th of Nokia's sales?

Pretty damned impressive, that.

That 1.5M number is way higher than the actual sales.

http://www.informationweek.com/blog/main/archives/2008/10/dont_buy_the_g1.html

(Note that URL truncates the full title, which is "Don't buy the G1 Sales Figures". It's not just an article telling you not to buy the G1)