Future Palmtop Computers

Comments on what should be in a future computer.

Pocket computers can and should work much better than existing industry standard computers based on Intel and Microsoft. The first requirement is not to buy one produced by people who have already proven that they can't provide reliable systems.

What Is Wrong With Industry Standard Computers?

The industry standard (IBM PC) carries multiple problems stemming from compatibility with the past, building down to a price, lack of support facilities, and very poor reliability. The original IBM PC, at its $2500 to $6000 price, and with a simple operating system was actually fairly reliable. The current problems with computers are mainly caused by buyers demanding low initial prices, usually without considering the long term hidden costs of using low reliability systems. Buyers value appearance above efficiency, performance above data protection, low price above reliability.

Computers can run for years without ever crashing. Many manufacturers routinely expect their systems to have a half hour of unscheduled down time per year. However the computers that do this tend not to be PCs. I supported a variety of computers for fifteen years. The various Unix systems from Sun, Silicon Graphics and Hewlett Packard got rebooted once every six months, whether they needed it or not. Once the Novell systems were moved from clone PCs to low end Hewlett Packard servers, they were rebooted once every six months.

Contrast this with your own experience with a standard office PC. Then ask yourself how much time you waste reinstalling and recovering after each crash.

I've listed below some specific PC problems future pocket computers should aim to avoid (and many present pocket computers do address and often solve these problems).

Too complex
Most non-technical people have problems with backups, corrupt registry entries, software and hardware upgrades, and system crashes. These were all taken care of by fulltime staff on mainframe systems. A good pocket computer will hide most of this complexity.
Data not protected
The value of a computer system lies mostly in the data it contains. Professional system had well planned systems for backing up and protecting data, facilities for moving data when machines were updated, kept multiple generations of data, and held backups off site. Data on professional systems may be retained on RAID style drives and be almost immune to hardware failure. See Plan 9 from Bell Labs for an example of how well backups could be done. A good pocket computer will automatically backup all its contents elsewhere whenever it connects. It will keep several generations of backups.
System not protected
A user can accidently delete system level resources, by formatting the C: drive, erasing executables or DLLs, etc. Professional operating systems have varying levels of protection, and make such accidents much less likely. Even some commodity operating systems, such as OpenDos, have flexible password protection for files. Windows 9x now protects critical system files, but since it is built on DOS, a user can still delete system resources. A pocket computer keeps most system files in ROM, which can not be altered.
Adding storage requires system changes
Professional systems such as Unix can flexibly add additional storage systems, by mounting new drives at any time. Pocket computers such as the Psion 5 automatically adapt to replacement of their reliable solid state drives, and automatically allow access to programs installed on them. In contrast, industry standard Windows systems require you to reinstall your software.
Resource scarcity
Hard drive space is inadequately monitored, as are ram resources. Replacing a hard drive is rarely a transparent operation (unlike those on a Psion 5). Industry standard machines now try to warn you of such conditions, but do not automatically accept new software on removable drives.
Hard to move a working system
Commodity machines offer no standard method of transferring your working file system to another machine when you upgrade. You need to uses some third party utility such as Ghost or DriveCopy. Many pocket computers can easily move your entire file system to another pocket computer.
Power supply vulnerable
Professional systems have uninteruptible power supplies. They don't crash every time the power line blips. Pocket computers are by nature running uninteruptible power supplies (from batteries), with a small backup battery.
System software upgrade problems
All systems attempt to provide foolproof software upgrade systems. Ask someone who uses each system how well they manage it. The Psion 5 installation system works far better than many used by Windows.
No instant boot, and unreliable
It is blindingly obvious that all systems should turn on instantly whenever you need them. Professional systems tend to avoid this need by being sufficiently reliable to run without crashing for months and sometimes years at a time. This is what happens with IBM mainframes, Novell Netware, QNX, Unix, etc. Industry standard Windows systems typically crash often. Pocket computers should be as reliable as mainframes. My current Psion 5 has had two problems for which I (needlessly) rebooted in the past 30 months.
Restart where you left off
All your programs should remain just how you left them when you turn on a system. OS/2 and Windows attempt this, but not all that cleanly. Most pocket computers do this very well.

Alternatives to Personal Computers

The obvious alternative is something like the Net Computer. Simple, cheap, networked all the time. Tell it who you are, or swipe a smart card, and it sets itself up the way you like. If these were available in every hotel room and every office, you might contemplate not needing to carry your own computer.

However, do you trust that the network infrastructure will always be working wherever you go? Also, it doesn't help while you are in transit.

Future Pocket Computers

My own overriding concerns are size, weight and battery life. A computer that you don't carry with you always is too much of a compromise, so I think pocket size, and weight under 200 grams, is the ideal. Battery life should be several working days. You do have to sacrifice some convenience to get these, but I'm not writing about future notebook computers.

We might be wanting heaps of facilities including but not limited to:

Lots of components are common to many functions. A microphone for recording, speech recognition, and telephone. A speaker for telephone, computer sounds, recorder output, and computer output. Each of these would probably use a digital to analog converter (and a DAC can be made to do ADC), which could therefore also be used for test equipment, battery condition, joystick equivalents and so on.

Form Factor

The 3Com Palm Pilot took a great chunk of the market. It had a few characteristics that were unusual. Accepted handwriting (but so did the earlier, larger Newton Message Pad). Small enough to fit in a pocket. Cheaper than many other units. Excellent easy connection to a Windows 95 PC.

A small form factor isn't enough by itself. Neither the credit card sized Franklin Rex nor the Timex Data Watch made it big. No data input while on the move.

The ideal form factor is a transformer. That's right. One of those children's toys that changes shape to suit the prevailing conditions. With lots of little add on pieces for special conditions or facilities. Designing this would obviously be very tricky, and I'm sure there would be lots of compromises. I suspect some of us will end up wearing our computer, as much as carrying it.


Sometimes sitting in a chair and typing is the fastest way to take notes, especially lengthy notes. People who think you can totally replace a keyboard with voice input have never worked in an open plan office, on a plane, in a library, or underwater.

Five possible options are a touch screen (for those who rarely type), a small well designed keyboard (Psion 5 shows what can be done), a chorded keyboard, an external add-on keyboard, or a virtual keyboard.

The touch screen is the cheap option, can handle handwriting or a Graffiti style hand alphabet, or present a qwerty, Dvorak or simplified keyboard. Good choice if you don't need to input much, and the Palm Pilot shows just how popular a choice it can be.

A well designed keyboard gives greater input speed, and would be the way I'd go, since I tend to do a lot of input. I'd rate the Psion 5 keyboard the best ever made, for its size (it is however very small). I absolutely guarantee no-one who touch types well will ever like a palmtop keyboard, but sometimes portability is more important.

Chorded keyboards can work real well, but are so unfamiliar that market perceptions would probably kill it. I've tried the Twiddler, which also incorporates a mouse.

An external keyboard should take the form of a keyboard adaptor, so you can use any conventional keyboard you can find. Conventional keyboards need a five volt supply, which is a right pain, since your adaptor may have to supply it (unless you use a passthrough to a conventional computer). Transfer to the PDA should be via IrDA or wireless like Bluetooth, so you don't need any fancy add-on cables to carry. So the adaptor would be a microprocessor doing IBM keyboard to IrDA conversion. You'd need the PDA to accept this input correctly.

A virtual keyboard scans the position of your fingers while you type on any available surface. Sensors may be very tricky.

Message pad

Shouldn't do less than a Newton. You sometimes absolutely need a way to draw sketches or diagrams.

Microcapsule electronic ink may offer a flexible display. www.eink.com are jointly developing it with Lucent, who contribute plastic transistors. Present versions refresh slowly, about 10 Hz, and take about 90 volts for very high contrast, about 10 volts for low contrast. Current is about 500nA per pixel. Joseph Jacobson expects 12mW for a 12 inch display eventually. Resolutions over 300 dpi are expected. The image should be retained with power removed. Xerox's Gyricon was based on a 20 year old idea of Nicholas K Sheridon. A 15cm by 15cm 1998 prototype is powered by a finger sized solar cell and is about as thick as a rubber glove. Displays of 220 dpi have been built.

Tape Recorder

If you have a microphone and a speaker, then there is no reason not to let the computer handle digital recordings. Most competent 1997 PDAs handle this, and the better ones (Psion 5 for example) have external controls for record and playback. You can (somewhat expensively) use PCMCIA cards or compact flash cards as tapes. There are some solid state tape recorders available that do the same, and some allow transfer of their recordings to a desktop PC for voice recognition.


The PIM shouldn't do less than Psion's Agenda application, which was about the most competent PDA as at 1998.

Alarm Clock

Any PDA that can't handle flexible alarms isn't much use. At least a half dozen, with a concept of days of the week, flexible working days and the like. Each alarm should specify why it is set, and have customised sounds, including recordings from the tape recorder. This is for the simple alarms.

The PDA PIM should be the one handling elaborate alarms, say for business trips, items due next decade, etc. Shouldn't need more than 2^16 alarms, but shouldn't have much other restrictions. PIM alarms would give access to all PIM facilities, notes and schedules.

Voice input

If there isn't enough horsepower for voice input, do voice recording, and have a voice to ascii converter package for a desktop PC, accepting the file on PCMCIA or compact flash (you don't really want to download a large voice file every time).

Flexible Power

You would want your computer to work well from regular batteries, rechargeables, a car cigarette lighter plug, regular mains power, solar cells, a hand generator, etc. AA or some other standard size (unless you like paying $100 for a rechargeable for your mobile phone, instead of $3 for a rechargeable AA).


If you support cellular reception, there is no real reason not to have full RF bandwidth access, AM/FM radio, CB, RF scanner facilities, and so on. I currently have an ISA card in my computer that receives from 500 kHz to 1300 Mhz, and is fully under computer control, and that is 1995 technology. You really can't get too much reception without a decent aerial, but no reason to make limits apart from that.

I'd think a small box (don't think PCMCIA would do it), own power supply, with a wireless connection to the computer, and computer supplying all the display and control functions. Like the Rosetta WinRadio, or Icom, but with an IrDA or Bluetooth connection rather than ISA or serial.


A PCMCIA card sized scanner? Why not? Take a look at the few pen type scanners that exist.


A small CCD camera can do still photography and video or picture phone. Still photos can be transmitted as either fax or by email as a GIF or JPEG file. Not much infrastructure required for this. Either built in or PCMCIA card based.

If you can get a $50 plug in camera (and a $50 printer) for a Game Boy (camera released 1998), then these needn't be too expensive (and one has appeared in 2000 for the 3Com Palm).


Consider a telephone. The old fashioned ones were pretty well designed. You could see the dial easily, and dial while listening to the phone. The cradle for the handset was raised to protect the "on hook" switch and stopped the connection being broken if you accidentally dropped it. The handset fitted comfortably in the hand.

Recent phones have had lousy design. Too thick to hold comfortably. The keypad in the handset, just at the time when every bank and company want you to use touch tone answers to questions on their voice mail (what an idiotic idea!) And the hook switch is vulnerable to accidental disconnects! Have their industrial designers lost all touch with their own history?

No reason a phone shouldn't be built into the PDA.

buy that?
pda list

ericlindsay.com -> palmtop -> futurepc

Contact webmaster