Cathode Tan - Games, Media and Geek Stuff
logo design by man bytes blog

Friday, May 16, 2008

Why XP And OLPC Is A Good Thing

Yeah, I know. I'm a bit surprised too. I'm going to defend Microsoft here. I'm going to defend them pretty hard, actually.

News that the OLPC will offer XP (check out the video) has been met with a kind of rabid fanboy reaction that is usually reserved for knee jerk comparisons between screenshots to determine if a 360 bloom effect is better than a PS3 bloom effect.

Here's Mike Arrington of TechCrunch:

There are no financial terms being disclosed, although it wouldn’t be dumb to assume that not only is the software being supplied for free, but Microsoft made a healthy donation to the organization as well. The last thing Microsoft wants is for anyone who’s computer literate to think that a world without Microsoft Windows is possible.

On the upside, though, the pain of having to deal with Windows crashes may make some of these kids excellent technical support people over time. They’d just get lazy with Linux being so stable all the time.

If it isn’t obvious from what I’ve written above, I’m not impressed. OLPC is in danger of becoming a celebrity cause rather than a real attempt to bridge the digital divide. My guess is Linux worked just fine as an operating system for these machines.
-- Poor Children Of The World No Longer Will Have To Struggle With Linux

First - let us note that he assumes Microsoft has made a "healthy donation" to an organization trying to give poor kids laptops and manages to turn that into a bad thing. As if OLPC will be more successful if they had less money.

Second, Mike here continues to repeat a common shortcut ... referring to the OLPC OS as Linux. While that's technically accurate - it runs a version of Red Hat, this version, called Sugar, runs a pretty experimental UI. While Sugar has some interesting design concepts and I'm sure was developed with a truckload of good intentions - the result is a bit of nightmare.

I should note here that I've actually used an OLPC. I didn't use it for very long. I didn't use it for very long because in general I found the OS to be horribly confusing and after trying out some of the apps, it seemed to have more or less crashed.

This is a pretty important distinction Mike and most of the Internet is overlooking. This not about XP versus Linux, per se, this is about XP versus Sugar. And if you asked me which one I'd rather have a poor kid in a developing country use, I will say XP. And I will say it repeatedly.

It's not Microsoft's fault that Sugar is a mess of theoretical design concepts that were never properly tested. OLPC did no usability testing, a move that UI guru Jakob Nielsen called "reckless". OLPC was even proud of this fact:

But XO developers defend their approach, which grew out of a core philosophy of the MIT Media Lab known as "demo or die." Researchers are encouraged to build new things, critique them, and then make improvements—rather than doing a lot of concept-testing up front. They're backed up by John Maeda, a user-interface design guru from the Media Lab who has been watching the XO development process from its beginnings. "They're using the Steve Jobs method," he says, referring to Apple's famous chief executive and design whiz. "You don't use focus groups. You just do it right."
-- The Face of the $100 Laptop

"Just do it right" makes for a great quote and all, but it's actually a pretty miserable development philosophy. And for the record, it is not Apple's philosophy. Here's is the first line from the first chapter of the first part of Apple's document "Apple Human Interface Guidelines":

The best way to make sure your product meets the needs of your target audience is to expose your designs to the scrutiny of your users.

And the second line:

Doing this during every phase of the design process can help reveal which features of your product work well and which need improvement.

Not doing these things will lead to things like shipping a web browser without a location bar, or things like this:

To that end, Sugar offers a simple technique for moving objects—a document, say, or an image—from one application to another. A student can pluck a photo off of a Web site by clicking on it and dragging it to the left side of the frame. Then, after she launches another activity on the display screen, she can click on the icon for the photo and drag it onto the screen. The drop-off spot on the frame is conceived as a "pocket" that the kids can use to carry around things they want to use later.
-- The Face of the $100 Laptop

Because drag, drop and a desktop metaphor have worked so horribly for decades. That someone approached this OS thinking that kids will not get dragging and dropping makes me wonder if they ever bothered to even meet a kid before starting. Most I've ever known take to new technology remarkably well in that freaky "my five year old speaks better French than you" kind of way.

And while this probably reads as indictment of Sugar, which it is, the real point here is that all of this "Microsoft against Linux" talk really should be more about a "Desktop based OS against whatever the hell Sugar is" kind of talk. Maybe if this was XP displacing Ubuntu running Open Office scenario, I could all like "MS is evil" or stuff.

But if OLPC is actually supposed to bridge developing countries to the modern world, then showing them the computer paradigm the modern world uses instead of someone's untested experiment makes perfect sense.

So let's stop with the hate. Not everything Microsoft touches withers and dies. Just because a kid grows up using XP doesn't mean that he won't discover Linux or OS X later.

In fact, he's more likely to be able to discover both of those by having used XP out of the gate instead of Sugar.

Thanks to Thomas for the post that got me thinking about all this in the first place. Also of note is 90% Of Everything's usability review of Sugar.


Thomas said...

After you gave me the URL to those UI articles, I did download a VM image of the Sugar interface, and you're right: it's gratuitously weird. Among other things, it seems almost mean to have such a poor task switcher on an intentionally CPU-weak computer.

If OLPC was worried about their free software cred, they could have gone over to Xubuntu or one of the Red Hat-derived XFCE distros.

But frankly, I think Krstic was right: the OS isn't really important to the project. If you accept the whole constructionist theory, you could ship out Palm TX handhelds with a Python interpreter and get the same results. None of these kids are going to hack the kernel anyway.

Josh said...

Oh yeah, I resisted posting the quote from the BusinessWeek article spouting off about how great it could be that the kids could use Python.

Let me get this right, you decided that kids can't use a normal OS, dumbed this one to oblivion - but you want me to be excited that some kid will learn how to hack with Python? Right.

But I think you're correct. XP, Linux, Sugar - these aren't the deciding factors here. It's the approach. It would be great if technology was the deciding factor - because it's a lot easier to solve.

Josh said...

And oh great, they're porting it: