NN/g Home AskTog Interaction Design Section Holistic Design and the PalmOne Tungsten T3
AskTog, August, 2004
This is a case study about a marvelous piece of hardware, backed up by a decent operating system, but which, nevertheless, made for an unpleasant initial user experience.
Why? Primarily because the handheld community is not taking a holistic approach to supporting their customers. This, in turn, is resulting in lost sales for the developers and retailers and a declining trend for handhelds in general.
Three weeks ago, I bought a pair of Palm Tungsten T3 handheld computers for my wife and myself.
It took me all three weeks, full-time, to get them just the way I wanted them. It should have taken me two days, and those should have been two days of fun, trying out new "toys" and deciding which to buy.
The Tungsten T3 is a wonderful machine, armed with a blazingly-fast 400 mhz processor and 568 megs of memory (including the 512 meg card I immediately added).
Those high-capacity innards were what were to get me in trouble. What does a boy do with all that memory? He fills it up!
Unfortunately, both the Palm and the surrounding third-party community are still set up to fill up the 8 meg Palm Pilots of only a few years ago.
Join my intensive (and fun!) lecture/ workshop course. Sign up now!
Interaction Design course: Go from zero to interaction designer in just three days.
User Experience Conference Website There's more than my course at an NN/g conference. You'll find a breadth of other specialized courses and networking opportunities that will put you and your company at the leading edge of the design curve.
The worst were the instructions that came in multiple files that were mixed in among the files to be installed, so you couldn't tell which were which. In at least one of those cases, I had to contact the publisher for instructions on how to separate out the instruction subfiles, so I knew which files to actually transfer to the Palm.
At least on the Mac, using OS X, the files would arrive with the wrong creator, so that the normal way of double-clicking for installation wouldn't work. The files had to be manually dragged into the Install folderafter you worked through which were the actual files.
Why the contradiction between the well-designed sites and the crude installation folders that result? The sites are merchants only. The software itself comes directly from the publishers. It's like going into a nicely-laid-out Macys and returning home with a badly-designed and documented blender. Macys had no control over the blender itselfor did they?
The merchants need to have their trade association (surely they have one) promulgate standards for handheld software "golden masters." This is really simple stuff:
• If there are more than two files involved, the files to be published need to be in one folder and the instructions in another, with a shortcut to the instructions in the root folder.
• If there's a Mac version, it needs to be tested on a Mac, to ensure that it actually works. If there's a Windows version, likewise.
• Instructions should include "what to do if double-clicking fails."
• This is the only recommendation that will take more than five minute's work: The handheld companies need to supply a really simple install widget that will enable developers to enter the names of the files that need to go to the handheld and enable the user, with a single double-click, to move them all. Right now, each must be moved separately. Even better would be to embed the files within the installer.
Unfortunately, many of the registration forms demand a UserID following this year's D'ohlt format: One's current email address. Since more and more people change their email address on a frequent basis in order to keep one step ahead of the spammers, this ensures that not only will this scheme do the merchant no good, but the user will not remember which email address they were using on that particular day.
Two other "gotchas": Some sites offer significant discounts on subsequent purchases. (Of course, they've adjusted their prices accordingly.) The "smart" way to buy multiple applications is to buy one, then see what the confirmation email has to offer. This kind of labyrinthine marketing ultimately loses sales, as people forget to return.
The final "gotcha" was handed to me by PalmGear. They offered to let me buy multiple copies of the same software, a real advantage to me since I was setting up a Palm both for myself and my wife. Then, they asked for a second palm user name, which I supplied. Though they charged me for multiple copies of several different applications, they only informed the publishers of one of the two supplied palm user names. For $120, I got $60 worth of software. They are off my list of merchants until I hear they have addressed the problem.
I found only one merchant that could handle multiple sales correctly. Most would only allow sales for one handheld at a time, and some of these would actually block you from returning for a second copy! They were attempting to block multiple "buy" clicks by setting a time-out, but failed to test whether the user had changed the handheld UserID for the second copy.
It's the job of Quality Assurance, the software testing people, to test "edge conditions." Multiple sales of a single software product is a rather obvious edge condition. In all these cases, someone in QA should have been at least asking the question and, in the case of PalmGear, which encourages multiple copy purchase, actually testing to find out if it worked.
• Don't play games with your users by tempting them with special discounts after the fact. Lower your initial prices and stop making users who need no further software feel bad about you because they were such suckers they bought everything at once.
• Make it easy to buy multiple copies of the same software.
As it is now, merchants are throwing money away with both hands
The first problem is that the "pipe" between the desktop computer and the handheld is pitifully slow for large files. We are warned that a single MP3 file can take up to 10 minutes to transfer.
The second problem is compatibility. My wife and I have two different brands/sizes of SD Cards. With one, we can transfer MP3s. With the other, we are fresh out of luck. It hangs, and there's no way to discover what is causing the problem. (The card otherwise works perfectly.)
Fortunately, with music, there's a lovely work-around: Put the SD Card in a USB reader connected to the desktop and drag the files directly into the Audio folder on the SD Card. Fast, clean, effective.
Unfortunately, most of the video and image applications do things to the file before uploading them, and many can't transfer directly to a card, but must go through the tiny Palm pipe.
Several applications I reviewed had a desktop component. These desktop applications were, in a word, weird. They were difficult to understand, difficult to use, and difficult to remember how to use, so that subsequent experiences were equally bad. I finally had to start writing my own terse user manuals, which I store along with each app.
These problems are all about to get much worse. The next generation of handhelds will have internal hard disks. Instead of a half-gig of memory, they will soon have 40 or 60 gigs of memory. How long will that take to fill up at 10 minutes per song? You do the math.
• Developers need to look at what each other are doing and begin to develop de facto standards.
• Developers need to test for learnability, usability, and memorability.
• Developers need to decouple indexing and categorizing of documents from the documents themselves. Let me organize the view of the data on the desktop where I have access to a large screen and a mouse, etc., but let me drag standard data files directly onto the SD Card, if I wish.
• PalmOne needs to fix the pipe. We're not talking USB1 vs. USB2 here. I was using only USB1 when writing directly to the SD Card. It was still hundreds of times faster.
The designers need to start acting holistically:
•Examine the entire experience of owning a handheld.
• Watch users as they struggle through the first few weeks of ownership.
• Interview users after six months of ownership and find out if they are still acquiring neat new software and, if not, what's holding them back.
• Form or join associations and formulate standards for every aspect of the handheld experience.
• Streamline systems, so that choosing, downloading, installing, paying for, learning, and using products is easy and fun.
Only in this way will handheld sales begin moving upward once more.
You curiously omitted specific mention of Palm in listing Who Should Do What.
As the Palm Desktop long ago acknowledged, loading new Apps is just one aspect of managing synchronized information between a handheld and a digital hub or docking station. To my mind, the problem is that PalmOne has seen fit to deprecate this overall interface, although you and I might agree that it's the lifeblood of the value proposition for a PDA.
Just as Apple claims to be integrating a "Sync" engine into Tiger, PalmOne should have an API, design rules for determining concurrency, backup status, any DRM, etc. Perhaps they'd even supply the tool that 3rd party apps use, with drag-n-drop loading of apps. Even we lowly EUL-ees might be able to spec files that we'd like maintained in parallel, with what archiving, etc.
Of course, Palm would do well to explicitly court the major app vendors on each platform to make sure that xCal, xPhoto, xAddress, xTunes (I've loaded Chinese tapes onto my T3 for quick study during idle moments), xMail and so on, all have optimized interfaces to the synch/update/archive/DRM tool. Having a dominant API would be a competitive advantage... I imagine much of the same grief exists in the PocketPC, Symbian and various PhoneOS worlds, too. Maybe they could even license it since the soft- and hard-ware teams are theoretically separate these days.
Just another 2 cents,
Case in point: I tried to update the T3 ROM, as advised on the the PalmOne website. They provide no method of determining whether you even need to do the update. Instead, they demand that you provide 20 megs of room on the Palm for the updater to do its job. In my case, that meant backing up the Palm by HotSyncing, then creating a new identity with no extra files, then HotSyncing the ROM updater, all as instructed by PalmOne.
After going through all of this, the 20 meg application immediately identified my ROM as already up to date, so my 30 minutes of work was in vain.
If only it were that simple. No, my adventures were really only beginning. When I continued with the PalmOne instructions to restore my original user name, all hell broke loose. Suddenly, HotSync was restoring programs deleted years ago. I soon discovered these programs were still hanging around in the backup folder because, apparently, HotSync only sort of syncs.
Worse, many newer applications that have worked around the limitations of HotSync by building their own sync engines were blissfully unaware that my Palm needed to be restored at all. Documents to Go was the worst. The only way I could restore the documents was to pretend to delete themeven though they were no longer there, then drag and drop them for a new upload.
By the time I was through, 2.5 hours had been wasted to discover that a ROM upgrade didn't need to be done, something that should have been handled by a simple Info request on the Palm, a capability not incorporated in the design.
Yes, Walt is absolutely correct. PalmOne or PalmSource, whichever one now has this responsibility, needs to get back to the task of designing software, not just pumping out hot new hardware. It was the magical combination of the two that made the Palm what it is, and failing to progress on the software front could sink them both.
If you are considering or already own a PalmOne Tungten T3, click to reach the companion article, Make Your Tungsten T3 Palm a Monster Machine.
Have a comment about this article? Send a message to Tog.
Previous AskTog Columns >
Contact Us: Bruce Tognazzini
Copyright Bruce Tognazzini. All Rights Reserved