Thursday, May 01, 2008

But Will They Get Out To Vote?





Political commentary hidden inside GTA IV. It's crude, overlong, and only funny in parts.

Lenovo: We Don't Get It





So here's an ad for the Lenovo X300. I'll wait here patiently while you watch it.

"No compromises." Except for industrial design, screen, keyboard, and operating system. It does have features you always need on an airplane -- like multiple USB ports and an optical drive. Oh wait, I use at most one USB device on my laptop except for at home (where I run it lid down with a Mac keyboard, which works as a USB hub because Apple isn't retarded). And I avoid even putting CDs or DVDs in a laptop in transit to save battery power.

I don't own an Air, and probably never will, but I understand exactly why some people would want one, oh and yeah, it's still the top selling piece of hardware at apple.com/store (ahead of entire categories of other Macs).

The Travellers



I loved this comic way back when. Sadly it's dated rather badly (too many references to movies we've happily forgotten, such as "Life Force"). I'm glad to see someone it's still around.

It's The Old Open Technology Trick, Chief



The Open Screen Project is supported by technology leaders, including Adobe, ARM, Chunghwa Telecom, Cisco, Intel, LG Electronics Inc., Marvell, Motorola, Nokia, NTT DoCoMo, Qualcomm, Samsung Electronics Co., Sony Ericsson, Toshiba and Verizon Wireless, and leading content providers, including BBC, MTV Networks, and NBC Universal, who want to deliver rich Web and video experiences, live and on-demand across a variety of devices.

From mx:EverythingFlex



So, a bunch of the Usual Suspects have decided to partner up with Adobe in an attempt to (depending on how you look at it) keep Flash/SWF/FLV dominant in the web video market or help make Flash/SWF/FLV relevant again in the web video market.

We're Number One!



If you look at what format most of the video on the web is in, it's WMV (or ASX or AVI or whatever -- Microsoft's video container format du jour). This is because most of the folks online use Windows and have no clue how to encode it so that other people can watch it conveniently.

If you look at the video format that most people on the web watch, it's FLV (Adobe's video container format, which is Flash's video container format). FLV is used for YouTube, Hulu, and abc.com. It's very popular because, frankly, Flash provides the best options for providing custom web players and is solidly cross-platform (if it plays on a Mac it plays on Windows and vice versa).

If you look at the video format people actually pay to watch, it's QuickTime (i.e. MOV, Apple's video container format). Aside from that fact that Apple is selling a buttload of QuickTime, while other folks are merely giving away stuff, there's a nice little twist in this.

QuickTime is not only the oldest digital video container format, it's also by far the most flexible, general-purpose, and well-engineered. Unlike the others it was designed from day one as capture, edit, and delivery format, and to handle arbitrary numbers and kinds of media tracks. A QuickTime MOV can contain multiple video tracks (e.g. for different bandwidths or resolutions), audio tracks (e.g. languages, directors' commentaries), text tracks (e.g. subtitles or index markup), interactive tracks (e.g. custom interfaces), and -- well -- anything else you can think of. And it's been this way since day one.

Recognizing this, after developing two complete generations of delivery-only video formats (MPEG, and MPEG2/3) that weren't flexible and didn't even include proper time-synchronization, the Motion Picture Expert Group went to Apple and asked to license QuickTime. The MPEG4 video container is, in fact, QuickTime. So H264 videos (which constitute a large number of WMV and FLV videos) are in fact QuickTime videos in disguise. Another consequence of this is that the QuickTime container format is, effectively, open and has been since Apple essentially handed it over to the MPEG folks.

Now, what does this announcement by Adobe mean? Well, the Usual Suspects are essentially a bunch of companies who'll sign onto any bandwagon that sells more hardware and chews more bandwidth (Cisco, Intel), technical incompetents (such as the BBC, who, despite being a public broadcaster, have managed to pick one non-cross-platform proprietary content delivery technology after another, and had to be dragged kicking and screaming to make their iPlayer work on Macs or in FireFox) and the "we hate Apple" club (NBC Universal who will cheerfully burn millions of dollars to spite iTunes). So we can guess what their motives are.

Thanks Goodness It's "Open"



First of all, you need to realize that FLV was the only "closed" container format. You can already write custom codecs for QuickTime and Windows Media, and their container formats are already public. This is why QuickTime can play WMVs compressed with supported codecs and vice versa. This is also why you can't open up an FLV in QuickTime, even if it's compressed with a codec QuickTime supports. So Adobe is actually playing catchup here.

As for the "benefits" to the consumer of FLV being open... Well, when Apple controlled QuickTime every QuickTime video would play on any machine with QuickTime installed. But lots of MPEG4 videos won't play in QuickTime (despite being, essentially, MOVs) because the open nature of MPEG4 allows folks to write their own codecs. (QuickTime already allows for third-party codecs, but most QuickTime developers seem to write cross-platform codecs.) So, in essence, Adobe is opening up FLV to be just as lousy an end-user experience as every other more open format (QuickTime and WMV have been "open" in the sense that FLV is being opened since day one).

Of course, FLV is already a lousy end-user experience for anyone not using a computer. Flash on the Wii won't play many (perhaps most) FLVs because one of the first things to go when Adobe/Macromedia streamlines Flash is codec support. (The same is true for QuickTime on the iPhone.)

So, in the end, what does the announcement mean? Well, I guess Adobe would prefer manufacturers to settle on their newly "open" FLV container instead of the already open MPEG4/QuickTime container, or the already pretty much open WMV container. Big surprise. A bunch of manufacturers have signed on, but whether they'll ship Flash-only devices or Flash-too devices isn't clear. If you put H264 hardware in your device, will you deliberately spite H264 that isn't wrapped in an FLV container? I doubt it: not even Adobe does that (you can pass QuickTime movies with H264 encoding to SWFs as if they were FLVs and they just work).

Of course, if Adobe is actually going to publish the FLV format, Apple and Microsoft can start transparently playing FLVs with supported codecs, eliminating the need for Flash to play back H264 FLVs. It also makes implementation of <video>foo.flv</video> in HTML5 easier (again, without necessarily requiring Flash). So, on the whole, I don't see how this is much of a win for Adobe or the general public in the short term, and it seems like it will probably hurt Adobe in the long term.

Wednesday, April 30, 2008

The Design of Everyday Things



"The Macintosh is the first computer interface worth criticizing." Alan Kay*


Today, Apple has made usability all but a household world, and you probably don't need to fight major battles in software development projects to have some kind of iterative usability testing budgeted into a project. This hasn't been the case for very long.

Shortly before the dot com bubble burst, I remember an article about the design of e-commerce sites in which a number of test subjects were sent to a large number of major e-commerce sites with instructions to buy a specific item sold on each one. I don't have the article or exact figures to hand, but as I recall, in 70% of cases the users could not figure out how to complete the transaction.

It's relatively easy to fix usability problems in software. As a software guy, it's pretty horrifying to bump into the world of atoms (versus bits, to use Nicholas Negroponte's excellent dichotomy) and discover that the whole usability idea hasn't sunk in too deep.

I've worked in usability on and off (mostly off) for nearly twenty years. I remember trying to get managers and partners at Andersen Consulting to read the Apple Human Interface guidelines, just to see that this kind of thing could actually be codified in a useful way. In 1995, more than ten years after the Mac was released, arguing for usability testing and design standards was still pretty radical in the world of IT.

Not many books can actually change one's life in a real way, but I think The Design of Everyday Things by Don Norman** is one such book. Read this book and it will quite likely change the way you think about everything. Even though I thought I was pretty savvy about UI design, it changed the way I saw the world.

The reason I bring up this whole topic is that I've been brought head-on into the world of bad design by my twin girls. Baby stuff, stuck in the world of atoms, is oh, so 1983. You may have read my rant about breast pumps, so I won't revisit that topic right now, but here are a few examples of staggeringly bad design I've been living with for the last five weeks:

  • Baby Sleepers (pyjamas for those who haven't had babies) often have single zippers running from the baby's neck down to the tip of her left foot, meaning you need to completely unzip the whole thing just to check or change a diaper.

  • Disposable diapers are designed in such a way that it's almost impossible to tell which way up or around they are in a dimly lit room (like the one you'll be changing them in at two in the morning). They're also folded with the tabs where the baby's behind will be, so that you need to stick your finger between the diaper and the baby's bottom to tease them out (in the dark at 3am). Why not fold them the other way?

  • Our baby bath (the highest rated we could find) is designed with a little hammock suspended over a small tub. Using the provided scoop to get water from the reservoir to the baby involves threading a narrow gap each time. Simply altering the shape of the tub (which would cost nothing had they thought of it) would eliminate this constant annoyance.

  • Our tandem stroller (also well-rated) is designed to accommodate our car seats (good idea) but switching it into accept car seat mode requires remembering which bits need to be pushed back or released, and they snap back out of position at the slightest provocation (so you can't just leave the stroller in its more useful mode). And, here's the kicker, the stroller has a basket for carrying stuff beneath the babies, but putting it into car seat mode makes it completely impossible to get anything in or out of the basket. Did even ONE person test this device before starting production?

  • A mixed case is our baby swing. Its safety harness is admirably well-designed for quick release (a good feature since the last thing you want is to have to fiddle around a baby you've just lulled to sleep) but fastening the harness is a little like learning a magic trick.

  • There's a remarkable lack of color coding options for things like baby bottles. This is very annoying if you have twins and need to prep a bunch of stuff in advance specifically for each twin such that you can figure out which bottle is for which baby at 3am.

  • Whoever designed the labels for marking up breast milk sachets didn't think to simply mark days of the week etc. on the sachets so all you'd have to do would be to tick a box. I know writing out a date on a plastic bag is something I love to do ten times a day while dead tired. And boy, reading that writing is going to be easy, I bet.

  • And finally, our bottle warmer (again highly rated) is designed so badly that Don Norman might have lavished his most sarcastic accolade on it: "it probably won an award". Or to borrow a phrase from Roy and H.G., "it's a sad joke". The principle is this: it has an element and some measuring cylinders. You place the baby's bottle above the element (in a socket) and pour in a measured amount of water, then press a button. The element boils the water and then shuts off when the water is gone (I assume the element gets too hot and that triggers the off switch).

  • So to warm a bottle (remember, you're doing this while dead tired, in the dark, and it's three in the morning) you need to measure out a certain amount of water into this stupid tube, pour it in, and then warm the bottle. Here's the kicker though: a good deal of heat comes off the element after it switches off, so the bottle temperature is highly variable based on how long you wait after this piece of junk switches itself off. If you're heating two bottles for some reason, the second will always get more heat than the first. And that's assuming you measured the water into the damn cylinders correctly.

  • This design is only slightly better than the infamous coffee pot (with the spout pointing over the handle) that graces the cover of The Design of Everyday Things ... but of course the coffee pot is an intentional joke. This thing is a product that people not only buy and use, but recommend. Go figure.


It's not just that this stuff is thoughtlessly designed (I won't say it's designed by morons or incompetents, it's more likely it's not designed at all, or without any kind of user testing), it's that customers aren't complaining blue murder about it. It's like the people who argued that DOS was easier to use than a Macintosh and then went back to editing AUTOEXEC.BAT to try and squeeze out another 2K of RAM so that their program would run -- the user-base is too accepting of garbage for vendors to feel pressure to improve their lousy products. Just like in 1983 with software.

* I've seen this quotation in various forms in a lot of places, but don't have a definitive source, so it's certainly paraphrased and may well be apocryphal.

** This book rates only two stars on Amazon.com while quite mediocre novels often score better. One legitimate criticism of the book is that it really doesn't offer any recipes for good design, merely ways of criticizing failed design and (less often) appreciating good design. It's quite clear based on twenty odd years of no-one else managing to provide such recipes (beyond a few useful lists of heuristics) that this is really hard, if not impossible. In any event, I think this book is utterly brilliant, but it won't tell you how to be a great designer.

Monday, April 28, 2008

Interstellar Navigation by Dead Reckoning



Aside: so much for YouTube. Lousy job of encoding and it lost my soundtrack. So I rustled up some quick code and voila.






My favorite science fiction writer wrote a novel in 1967, give or take, called The Killing Machine in which the hero is attempting to locate a planet without precise coordinates, but has the assistance of a native of that planet who can remember the constellations of the night sky. The hero gets to what he believes is the correct "region" of space and asks her to look for a familiar constellation, because that will be direction to the star system in question.

This may seem far-fetched, but it's actually extremely practical (insofar as anything involving interstellar travel is practical).

I've never really looked into the technique in detail, but there's some discussion right now of a clue that the obsessive fans of Battlestar Galactica (the new series of course) who have noticed in the latest episode The Ties That Bind the constellation Orion showing up in several space shots. This would indicate that they must be getting close to Earth...

Now some folks have dismissed this, saying that it's bad science and they'd need to be incredibly close to Earth to recognize any constellations. These folks are just dead wrong. The two brightest stars in Orion are Betelgeuse and Rigel which are extremely bright and distant stars. The belt, similarly, is composed of distant stars. Orion would be easy to recognise from a considerable distance -- if you were looking straight towards Sol.

The video I've attached demonstrates all this using the wonderful free program Celestia. The upshot: Orion is easily recognizable from over one hundred light years out from Sol (assuming you're on the far side of Sol from Orion) and every other constellation is distorted beyond recognition.

Sunday, April 27, 2008

Breast Pumps are a Racket





My wife is using a Medela breast pump in her ongoing efforts to avoid raising our twins on formula. We actually bought a Medela breast pump before the twins were born, but the hospital gave us an even more up-market one as a "gift" so we ended up returning the one we bought.

First of all let me say this. Breast pumps are a racket. You can buy a pretty robust device for pumping up car tires that runs on 12v for $20 and it's probably better constructed than a typical breast pump. I'm sure there are lots of considerations that drive up the cost of a breast pump compared with a dirt cheap tire pump, but come on: The Medela breast pump we bought retails for ~$300 and is approximately as well constructed as a decent quality toy. OK it's made of non-toxic plastic, and it presumably is designed not to rip a woman's nipples off by accident, but seriously.

Our breast pump has a "valve" designed to allow milk to come into the collection bottle but prevent air from leaving the bottle when the pump "sucks". This "valve" is a piece of flexible plastic the size of a dime weighing a fraction of a gram, and it costs $5 for two of them. (So far we've lost three down the drain.) This is a freaking scandal.

It's not like breast pumps are so uncommon that economies of scale don't exist. Our hospital is giving a breast pump to every other woman who gives birth (it's that or a stroller; if you do the math, you take the breast pump). Everyone gets born. The average woman gives birth two and half times. What. The. Frack.

Now, get this: the best approximation for the shape of a woman's breast that the genius designers at Medela can come up with is a cone. They sell big cones and small cones. Ameda (actually produced by the company formed by the guy who invented the electric breast pump) offers a silicone widget that's designed to simulate a baby's mouth. But apparently Ameda hasn't figured out that by selling their breast pumps for less, people assume they're not as good. (I can't find any review sites that indicate Medela are as well-liked by their users as Ameda, despite the "Stockholm Syndrome" that anyone who buys a more expensive product tends to suffer from -- yes Apple we're talking about you.) Any site I've found which shows reviews of both Ameda and Medela products, the Ameda products (which are cheaper) get better reviews.

Again, Ameda have done a bunch of really decent stupid things such as (a) making their gear completely compatible with third party bottles, (b) making their simulated baby mouth widgets compatible with third party breast pumps, and (c) providing excellent customer service (according to numerous reviews I've read). What they obviously should have done is sell tiny plastic flaps for $2.50 (Ameda's valves are larger and won't wash down sinks) and force users of their pumps to buy their expensive nipples and collars and single-use-leaky-plastic milk storage bags which would make their products more profitable and allow them to hire reps to convince hospitals to use their products in more hospitals, and be able to give away their laughably overpriced pumps to new moms.

We've probably spent about $100 buying plastic doodads compatible with our "free" breast pump so far, but after losing a day's worth of milk to their lousy (and expensive) storage bags, and discovering today that (unlike Ameda) we need to sterilize the plastic tubes (joining the pump to the collection gizmo) of our Medela system if moisture gets in (Ameda's pumps are completely isolated from the collection system by means of a local cylinder) we're on the verge of setting aside the entire system and switching to Ameda.

In the end, as I said to my wife, it comes down to this. If you have to use a machine to suck on your breasts, would you prefer one designed by Swedes or Germans?