> MS invested a lot in PDAs, at a time when the tech just wasn’t good enough.
And they were not at all alone in that. A lot of us were totally blinded by that wave of PDAs and first experiments with tablets. E.g. I spent '98-'99 largely working on a tablet [1]. It was a beautiful piece of hardware for what we could do with the constraints, but those tradeoffs also all reflect the limitations that killed it:
The back "tube" held the batteries and doubled to lift the back slightly to make it easier to read the screen if you put it on a desk like a phone. But you could not get away from the fact it made it builky. It used a DECT (cordless phone standard) extension because wifi was not yet common, but of course the moment wifi took off nobody wanted that, and it had to get wifi.
We spent lots of effort writing a GUI toolkit for Nano-X (an compact little GUI engine with a somewhat X11 and somewhat Win32 API frontends [2] ), and contributing a bunch of code for Nano-X itself. And while Nano-X was excellent for what it was, all of that effort was down to memory and flash limitations (32MB RAM, 16MB flash), and meant applications had to be custom written, as while the API was X11-like it was not a direct match. And browser based apps was certainly not a option because that, well, it wasn't really much of a thing yet.
We spent ages with Opera getting them to port Opera to work under those constraints (to their credit the browser engine took very little time to retarget), though, and the results were great for the size of the device, but again it was a choice that was expensive for a small startup, but necessary because the device was so hardware-limited.
The bezels are massive partly because a bigger screen would have been too expensive, heavy and power hungry, partly to accomodate the rest of the board.
Touch was all resistive, because capacitative touch screens where not a viable option at the time.
The batteries didn't last long. Not that it mattered much, as you'd have to stay within a few hundred meters of your DECT base-station anyway, and so couldn't really take it out of the house.
And many more.
That it'd fail to get traction seems obvious now - it got geeks excited (especially because it ran Linux). I loved working on it. But just like "everyone" I knew had a Palm Pilot but Palm Pilots still didn't break through into the mainstream, this didn't get regular people excited either, and unlike Palm Pilot it didn't fit into a niche that had already gotten itself semi-established through a long range of previous devices (e.g. Palm Pilots gave mobile computing; FreePad didn't, and was too bulky anyway). We were not the only ones by far - e.g. Ericsson had a project to produce a Linux "screen phone". A bunch of others tried laptops with rotatable screens, or more directly tablet-like devices.
So when all of the various attempt to make the PDA and tablet markets break into the mainstream failed, and the iPhone then came, a lot of us looked at it with the eyes of someone who had seen it all before and we didn't spot what made it different:
This was suddenly a mass market consumer device rather than a niche techie device. We compared it to J2ME phones and complained about too few features, rather than comparing it to the vision of what we had wanted our devices to be (which was largely Star Trek or 2001 inspired - don't let anyone suggest this class of devices were invented this recently; we never thought the idea of that type of device was new) and spot that they'd done that, but had overcome enough of the worst limitations that the hardware constraints of the first wave of these devices had to deal with and designed something that would appeal on look and feel and capabilities to end-users, rather than appeal to super-early adopters.
To me it's one of those things that have made me much more cautious about judging new tech. Especially when I catch myself thinking "that's not new" I try to force myself to look for what is new about it, because it's so easy to judge something based on flaws of past implementations rather than by what the new iteration does differently.
[I just noticed at least some of the phone numbers on the dialler screenshot in the referenced article appears to be real; and most of the people listed were working on the project at the time, including me]
[1] http://linuxdevices.org/freepad-norways-alternative-to-swede...
[2] http://microwindows.org/