DocBug

Unbranding an unlocked Cingular/AT&T Treo 650 

My Treo 600 has never had good voice quality and the past week or two has started crashing when wireless is used, so I picked up a 2-month-old Cingular/AT&T Treo 650 off of Craig's List. In the US, smart phones are mostly sold by carriers who are hellbent on locking their customers into their own revenue streams, which means most cellphones on the market are "locked" to a particular carrier and some of the functionality, like the ability to use the cellphone as a wireless modem over bluetooth, are deliberately crippled. [Update: apparently, both Cingular and Sprint have now uncrippled this feature — good for them.] The previous owner of my phone had already finagled an unlock code, so my phone can be used with any GSM carrier just by inserting a new SIM card, but it was still running the Cingular-branded ROM. This weekend's project has been to replace it with the generic unbranded Palm-OS ROM — which turned out to be much more difficult than I had expected. Below are some notes on the process, intended mostly for people searching Google after running into the same troubles I did.

Note: All the instruction sheets I've run into warn that trying to upgrade your firmware to the generic version can be difficult and may very well ruin your Treo completely. Attempt at your own risk.

I started by following the instructions at Uneasy Silence, with help from a description at Treonauts. The upgrade is a two-step process: first you need to upgrade to the beta version of the unbranded firmware & software (1.23 & 1.06), and then to the current 1.28/1.13 versions. That's because all versions since 1.23 now check to see if you've got a branded ROM already, and refuse to upgrade if you do.

The first attempt to upgrade to 1.23 ran through the steps but then ended with an error message, leaving me in an odd indeterminate state. My phone's firmware showed up as 1.23 and the software version showed "Treo650-1.06/2078-..." with the last part of the version not fitting onto the info screen. I think the missing part was ROW, the ID for the unbranded-Treo version of the OS, but the phone still showed the Cingular logo on reset. The phone was also somewhat flakey in this state, with hotsync not working from the cable and other attempts to upgrade to 1.28 resulting in the cryptic error ""Different partition detected. Hotsync installer cannot be used. To upgrade your device, use the SD installer." It took me a while to recognize that the upgrade had only partially succeeded and to redo the 1.23 upgrade process a second time, but once I did the phone showed the generic Treo logo on reset.

Upgrading to the 1.28 firmware and 1.13 software was harder, and I still have no idea what the problem was. If I were running Windows I'd just run an EXE file, but since I'm using Mac OSX the official instructions say to hotsync to load 44 files, at which point the upgrade program would run. I forgot to bring my hotsync cable with me for Thanksgiving break, so I tried hotsyncing via Bluetooth instead. Unfortunately, the phone continued to reboot in the middle of the hotsync, try running the update software with incomplete files, then complain "There is not enough memory available to complete this operation." (I suspect this error message is spurious — it came up in all sorts of situations.) Copying the 44 files onto an SD card, transferring them to the phone using Launcher X and then doing a soft reset led to the same error, as did transferring only the DB files over using the SD card and using the normal Bluetooth hotsync to transfer the program files. Running the "DeviceConfiguration" program directly from SD card gives an error about not being able to find its scenario files.

What eventually worked was this:

1. Copy the 44 files onto my SD card
2. Do a hard reset to the phone, thus erasing all files
3. Beamed Launcher X from my Treo 600 to my 650
4. Inserted the SD card into the 650, and used Launcher X to copy the 44 files from the SD card to RAM (handheld)
5. Without doing a soft reset or removing the SD card, I ran the program "D.C." which now showed up in the applications section
6. Plugged the phone into power, pressed OK and watched the upgrade (finally) go off without a hitch!

If this weren't a new phone, at this point I'd have used hotsync or BackupBuddy to restore all my data to the phone (you did make a backup before starting, right?).

I have no idea why this worked when nothing else did — for all I know the planets were just in alignment this time and my particular method didn't matter one bit. If this helps or works for you, great! Let me know in comments! If if doesn't work feel free to leave comments too — I won't be able to help, but I'll a least sympathize.

Update 11/26/05: After searching around a little more I see that Sprint and Cingular both used to cripple Dial-Up Networking (DUN), but apparently Cingular enabled it in February and Sprint enabled it in June.

Very fly 

I picked up a Fly Pentop the other to play with (one of the advantages of being a user interface researcher is all the toys :). Here're a few thoughts.

There're at least four challenges with using a real ink pen as a computer interface:

1. There's no way to constrain where the user draws: strokes that should be invalid in your application leave marks just the same as valid strokes.
2. There's no way to tap or "drag" without leaving a mark, and no way to erase marks after you've left them.
3. There's no display to give feedback to your actions or computation. (The Fly uses audio for feedback, which gets around this limitation to some extent.)
4. You need a pre-printed sheet of paper for every pre-designed interface — the Fly comes with about 15 pre-printed games sheets, everything from word-search pages to maps you can tap on to test your geography knowledge. That doesn't scale well when dealing with applications with many different pages or when maintaining large numbers of applications, especially given that pages get "used up" when you write on them. On the other hand, the fact that paper is consumable makes for market opportunities that normal software doesn't have... I expect Leapfrog isn't too upset about that fact.
5. If you don't use a pre-designed interface (that is, you start with a blank sheet of paper) then the user is forced to draw the entire interface. The Fly has one game where you draw a piano keyboard that you can then play by tapping on the keys, but first it gives you explicit instructions like "draw 9 short vertical lines, going from left to right" to make sure you draw a keyboard it can understand.

In spite of these limitations, it's extremely engaging to be able to draw your own functional user interface — as anyone who read Harold and the Purple Crayon or watched Simon in the Land of Chalk Drawings as a kid knows. The effect really hit me when I was making a calculator. First I wrote the letter "C" and circled it to enter calculator mode. Then, as the pen spoke instructions to me, I drew a big rectangle and started to fill it with numbers and arithmetic symbols. I realized about three numbers in that I didn't have to stick to the usual layout and placed the rest of the numbers going up, down and sideways. Then I tapped on the numbers with the pen to type out 22 + 44..." only to discover I'd forgotten to draw an equals sign. I quickly drew one in, then tapped it to hear the pen speak "22 + 44 equals 66". It was as if I were running from something in the land of chalk drawings and someone suggested we draw a door so we could escape!

The interface also feels more magical than it would if it were implemented on a tablet PC. This could be a novelty effect — I'm used to paper being static and non-functional and computer screens being reactive — but I think it's also because it feels like the pen is reacting to my physical environment, rather than simply reacting to the way I interact with it. When I interact with a tablet PC, I think of the computer as being the screen (even if the actual CPU is somewhere else). With the Fly, I think of the pen and speaker as being the device, but not the paper. That means even though a tablet PC and the pentop computer might implement the exact same interface, I feel more of an emotional attachment with the pen because it appears to be observing and sharing my external environment and not just the actions I perform directly on the device.

Fly Pentop Computer 

Image: Leapfrog

The NYTimes has a write-up on Leapfrog's Fly Pentop Computer, which essentially merges the Anoto Pen technology with a speaker and what sounds like some very clever games & applications, all wrapped in a $100 pen. Supposedly it's for the 8 to 14-year-old market, but I'm thinking it might be good for this 30-somethinger as well.

(Thanks to Ted for the link.)

Update 11/25/05: I picked one up at Fry's a couple days ago — here're some thoughts on it.

Colored Bubbles! 

Ooh! Popular Science reports on a new invention: soap to blow colored bubbles that don't stain your clothing (or the walls, or the dog):

Tim Kehoe has stained the whites of his eyes deep blue. He's also stained his face, his car, several bathtubs and a few dozen children. He's had to evacuate his family because he filled the house with noxious fumes. He's ruined every kitchen he's ever had. Kehoe, a 35-year-old toy inventor from St. Paul, Minnesota, has done all this in an effort to make real an idea he had more than 10 years ago, one he's been told repeatedly cannot be realized: a colored bubble.

(Thanks to Ricky for the link!)

Using context to suggest recipients for a photo 

Marc Davis and others working at UC Berkeley's Garage Cinema Research group have some interesting work on using a person's context when taking a photo with a cellphone (specifically time, location and people who are around) to predict who that photo is likely to be sent to [paper, video]. They're using that prediction to offer a "one-click" list of people with whom to share a photo that's just been taken, and report that 70% of the time the correct sharing recipients are within the top 7 people listed. In their study, they found that time was the best predictor of who a likely recipient would be, even beating out what other people were around (determined by detecting other cellphones in the area via Bluetooth).

It's interesting to compare this to my own work [paper] using the Remembrance Agent on a wearable computer, where I found relatively little benefit in using either location or people in the area to suggest notes I had taken in previous conversations that might be useful in the new situation. It's clear that the application and user's lifestyle makes a huge difference. All my notes were taken when I was a grad student, so over a third of my notes were taken in just one of three locations: my office, the room just outside my office and the main classroom at the Media Lab. That's too clumped to help distinguish among the wide variety of topics I'd talk about in those locations. On the other hand, people in the area had the reverse problem: since I'd be giving demos and talks all the time, over a third of the people I was with when taking notes showed up only once. The "people who are around" feature was too sparse to be helpful. (I never did test time-of-day or day-of-week as feature vectors, because I dropped that feature from the RA when I wrote version 2, but I suspect it would have the same problem location does.)

Choice Blindness 

One of my favorite psych studies is one where a subject who had had his left and right brain hemispheres severed was asked to point (with either his left or his right hand) to one of four given pictures that matches a test picture. Unbeknownst to the subject, his right eye is shown one test picture (say, of a chicken claw) and their left eye is shown a different one (say, a snow scene). When asked to point with his right hand to the matching picture he picked a chicken, when asked to point with his left he picked the snow shovel. The fascinating part is when the subject was asked to verbally explain why he picked the snow shovel. Language is mostly generated in the left hemisphere (which controls the right hand), the half that didn't pick the shovel. Rather than look confused, he invariably came up with explanations for why he picked what he did — explanations that the experimenter knew were incorrect like "oh, you need the shovel to clean up the chicken coop."

Now BPS Research Digest points to a new study where they find the same sort of "choice blindness" in normal subjects:

One hundred and twenty participants were shown 15 pairs of female faces (taken from here). For each pair they had to say which of the two faces they found more attractive, and on a fraction of trials they had to say why they’d made that choice, in which case the photo of the face they’d selected was slid across the table to them so they could look at it while they explained their choice. Crucially, on a minority of these trials, the researchers used sleight of hand to surreptitiously pass the participant the photo of the face they had just rejected, rather than the one they’d chosen.

Bizarrely, only about a quarter of these trick trials were noticed by participants, despite the fact the two faces in a pair often bore little resemblance to one another. Even stranger was the way the participants then went on to justify choosing the face on the card they were holding, even though it was actually the face they’d rejected. It’s not that participants weren’t paying attention to the face they’d been passed – the justifications they gave often related to features specific to this face, not the one they’d actually chosen. Independent raters who compared participants’ verbal explanations for choices they had made (non-trick trials), with their explanations for the choices they hadn’t made (trick trials), found no differences in amount of emotional engagement, degree of detail given, or confidence.

As I've said before: Man is not a rational creature. Man is a rationalizing creature.

And the first OSX Trojan in the wild goes to... Sony? 

Last year there was a bunch of hoopla about the "first Mac OS X Trojan Horse", a claim that was quickly dismissed as being a non-issue since it was just a proof of concept and wasn't found in the wild.

Now it looks like we may have the first real Trojan for OS X found in the wild... being distributed by Sony. According to a tip published in Macintouch (and reported in The Register), Sony BMG is is including Mac-aware DRM software from Sunncomm in their new release of Imogen Heap's CD Speak for Yourself. The application, innocuously called Start.app, installs two kernel extensions that implement Suncomm's DRM scheme.

In their defense (legal, if not moral) the software does pop up an End User License Agreement that tells you what they're going to do — and I'm sure you all read those EULAs in their entirety before clicking OK, right?

Clay Shirky talk on "Making Digital Durable," Nov. 14th in SF 

For you Bay Area locals, this month's Long Now Foundation talk is Clay Shirky on Making Digital Durable: What Time Does to Categories. Monday, Nov 14th at 7:30pm at the Cowell Theater at Fort Mason (in San Francisco). A $10 donation is requested but not required.

Frappr 

Frappr is a simple cross between flickr and Google Maps: create a group and members (actually, anyone) can post a picture of themselves associated with their zipcode. Simple idea, and quite cute.

For an example, may I suggest the new Docbug Reader's group :).

(Thanks to Del for the link.)

iPod-integrated Headphones 

As I mentioned a few weeks ago, the iPod Nano is small enough that the headphones are becoming the limiting factor for size. Now Macally is drawing the obvious design conclusion with their mTune Chordless Stereo Headset by integrating the Nano right into the headphones.

(Thanks to Rawhide for the link.)

Functional PowerShot S200 halloween costume 

Winner of Engadget's Halloween Costume Contest: a functional Canon PowerShot S200 costume, with working rear LCD viewfinder. (Thanks to Nerfduck for the link.)

5000 years of prior art is "limited"? 

The US Patent & Trademark Office just published an application for a patent on a particular storyline, filed by rocket-scientist-turned-patent-agent Andrew Knight [Register article]. If granted (a big if) this could open literary plots to patent protection much like the 1980 Supreme Court case Diamond vs. Diehr opened the way for patents on software and algorithms.

I'll leave the debate on the legal points to the experts, but I can't help but imagine all the engineers in the Patent Office trying to decide whether Knight's plot idea is both novel and non-obvious to one well-versed in the field of story-writing. In a way, the situation is very much like 1980, when the USPTO wouldn't hire computer scientists to evaluate software patents because they only hired people with "Engineering" in their degree. Sure, the evaluators would catch if someone tries to patent the plot to The Matrix: Reloaded, but will they recognize a plot from an obscure Henry James novel, or one only published in Chinese?

The significance of this lies in how prior art is handled differently by copyright and patent law. Plots are currently covered under copyright law, though unlike patent law there are no "claims" that are written out in advance to specify just how different a new work must be before it no longer infringes. The key difference for prior art is that copyright only protects actual copying of an expression of a work. If you write a screenplay and then I produce an almost identical screenplay without your permission, that's only illegal if I actually copied your work. If I can convince a jury that I'd never even seen your work and we just happened to come up with the same idea at the same time, I'm off the hook. In fact, in that situation both of us could hold copyrights on our respective versions, regardless of their similarity. Prior art in copyright cases is really just two simple questions of fact: did you come up with your work yourself (i.e. not copy from anyone else), and if so then did I copy your work?

Patents, on the other hand, can be infringed whether or not the infringer knows the patent exists. If you and I both independently create the same invention (which happens more often than you'd think) then whoever filed for the patent first (or in the US, whoever invented first) gets exclusive right to the invention. The flip-side is also true: if the process you are trying to patent has already been published then you can't patent it, even if you never knew about the prior art. For both parties to a suit, copyright treats originality as anything that came from the author's head, while patent law treats originality as anything that is literally new under the sun.

Knight's page discussing the expected value of a storyline-patent application emphasizes the prior-art problem with a rather chilling assertion:

The breadth of Storyline Patent protection available, before the Federal Circuit approves Storyline Patents and the floodgates to the Patent Office open, is far greater than after. Beat your competitors to the Patent Office and be among the first applicants to enjoy examination over limited prior art. [Emphasis mine]

From that quote you might think that the only prior art checked for a patent is that which already exists in the patent database. That may even be true in practice, due to examiners' limited expertise and resources, but as I already pointed out, for patents prior art includes anything that has ever been published or disclosed. Ever. If you're patenting a new automobile engine then that's a relatively bounded problem — the gas engine was invented less than 150 years ago, and the field of engine design is relatively small. Other technical fields are similarly bounded because technological innovation tends to be incremental, building on top of prior inventions like towers extending into a downtown skyline.

Stories, on the other hand, can emerge from anything connected to human experience. If new technology is built on top of the old like a densely-packed downtown, stories are like wind-carried seeds that can take root almost anywhere. This is not to say that every story is unique — on the contrary, after 5000+ years the landscape is fairly crowded. Sure you can set your story of star-crossed lovers on the West Side of New York instead of Verona, but the difference between your new plot and prior art (and thus what your claims could cover) will be pretty narrow. But that's only true if someone can actually find the particular stories that read on your claims.

I suspect (and certainly hope) that this application will be rejected and the whole question will be moot, but I could very well be wrong: the trend for the last 25 years or so has been towards granting more exclusive rights to content holders, not fewer. Luckily, I also don't expect a disaster even if we do see storyline patents in our future. Precisely because stories are such a broad field, a 20-year limited monopoly just doesn't have the same detrimental effect on innovation like a patent on a key technology can. And compared to the lifetime of an author plus 70 years (plus an additional 20 years every 20 years), a single 20-year government restriction on the free market of ideas sounds like a bargain.

Remember, remember 

Remember, remember the fifth of November
Gunpowder treason and plot.
I see no reason why gunpowder treason
Should ever be forgot.

Guy Fawkes, Guy Fawkes, twas his intent
To blow up king and parliament.
Three score barrels were laid below
To prove old England's overthrow.

By God's mercy he was catch'd
With a dark lantern and lighted match.
Holloa boys, holloa boys, make the bells ring
Holloa boys, holloa boys, God save the King.
Hip hip hoorah!

A penny loaf to feed the Pope.
A farthing o' cheese to choke him.
A pint of beer to rinse it down.
A faggot of sticks to burn him.

Burn him in a tub of tar.
Burn him like a blazing star.
Burn his body from his head.
Then we'll say ol' Pope is dead.
Hip hip hoorah!
Hip hip hoorah!

(That would be 400 years ago, today...)

Amazon harvesting natural intelligence 

By way of John Battelle's SearchBlog:

Amazon Mechanical Turk provides a web services API for computers to integrate "artificial, artificial intelligence" directly into their processing by making requests of humans. Developers use the Amazon Mechanical Turk web services API to submit tasks to the Amazon Mechanical Turk web site, approve completed tasks, and incorporate the answers into their software applications. To the application, the transaction looks very much like any remote procedure call: the application sends the request, and the service returns the results. In reality, a network of humans fuels this artificial, artificial intelligence by coming to the web site, searching for and completing tasks, and receiving payment for their work.

The service is legit — http://mturk.amazon.com/ redirects to the main site.

On first blush it sounds similar to OpenMind (which was started by David Stork, a coworker of mine). OpenMind has especially been used to gather human knowledge for training up AIs (especially common-sense knowledge) — I wonder where Amazon expects to go with the idea.

Google Print 

Yesterday Google announced that the first wave of Google Print is up, starting with a large number of books that are in the public domain. (As they say in their PS, see also The Million Book Project's libraries in the US, China and India as well as Project Gutenberg.)

Demo of xMax 1000 times more efficient than WiMax 

Wow. Techworld is reporting on a demonstration of wireless communications sent at 3.7Mbit/s to a radius of 18 miles using just 50mW and an omnidirectional antenna using a technology called xMax, developed by xG Technology. If this is for real, that's on the order of 1000 times more efficient than GSM, CDMA or WiMax. The company plans to target long-range wireless, but Princeton EE professor Stuart Schwartz claims he has seen it also demonstrated as a personal-area network, giving 2Mbit/s over 40 feet using just 3 nanoWatts.

If this is all true then it's revolutionary. To his great credit, Techworld reporter Peter Judge has a full companion article laying out the several places where reporters have to take the company at its word about the technology and the honesty of the demo, as well as remaining potential hurdles such as preemptive regulation and the possibility of reflections or interference once other transmitters start using the same system. But we'll know soon enough whether it's more than just snake oil, and if so it's going to be darned impressive.

(Thanks to Kurt for the link.)

Understanding science in the popular press 

Ben Goldacre's latest Guardian column Bad Science has a nice look at how journalists fail to correctly interpret scientific studies, in particular looking at a recent flap over the (lack of a) link found between autism and vaccines. His concluding advice to science journalists and presumably the rest of us:

...if you don't get it then you have only two choices: you can either learn to interpret data yourself and come to your own informed conclusions; or you decide who to trust. Choose wisely.

(By way of Mind Hacks.)

Sony BMG installing rootkits when you play a CD 

So what marketing genius over at Sony BMG decided that rootkitting the PC of anyone who plays their CDs is a good way to attract customers?

Every time I think the recording industry is going to get a clue, they just go further off the deep end. Next thing you know they'll be secretly bugging our bathrooms to keep us from illegally singing "their" songs in the shower...

Update 11/2/05: Declan McCullagh over at Politech asks an interesting question: Does this rootkit constitute a "technological measure that effectively controls access to a [copyrighted] work?" If so, would removing the rootkit from your own system constitute a violation of the DMCA?