Why Bluetooth Sucks
While researching my idea for a Skype Plugin for Bluetooth Headsets, I came to realize just how much Bluetooth sucks. As a technology, it seems reasonable enough, albeit as complex and elaborately generalized as one would expect for a technology that emerged fully formed from a standards body.
However, it takes more than technology to yield a technology that doesn’t suck.
On PC side, you need a Bluetooth radio, obviously, but you also need a Bluetooth ’stack’. This is akin to the third-party TCP/IP stacks you used to have to buy to use IP networks, back in the dark ages. There are competing vendors, of course, including Microsoft’s crippled BT stack in XP SP2, the dominant Widcomm stack, and others. Of course, there’s no common API like the socket API shared by TCP/IP stacks, so you have to pick one and code against it.
The APIs are not open (save for MS’s), so to code for the Widcomm stack you buy the Widcomm Windows development kit, to the tune of USD1400. The apps you produce with said stack won’t work with non-Widcomm drivers. Awesome.
Now, Widcomm had an opportunity to drive Bluetooth adoption by providing a free development toolkit for interfacing with Widcomm Bluetooth drivers, and selling the driver suite to hardware manufacturers, who notoriously suck as writing their own drivers. Instead, Widcomm opted for the short-term revenue-maximization approach typical of an Old Economy hardware vendor.
Now, if I want to write an Skype plugin which would increase the value of a Bluetooth headset for anyone who uses Skype, I have to shell out for at least the Widcomm development kit (since most BT hardware uses Widcomm), but even then I won’t be fully covered, since there are BT devices that don’t use the Widcomm stack.
Recall that it was the standard, cross-platform sockets API that allowed tools like FTP, gopher, telnet, and eventually Mosaic to be written once and used on a wide range of TCP/IP stacks. Imagine if Mosaic was tightly coupled to a specific vendor’s stack (say, NetManage), and Andreesen et al had to shell out $1500 for the privilege of coding to the NetManage API. It’s like that now with Bluetooth, and it sucks.
Will Microsoft please write a complete Bluetooth stack, maybe even incorporating the Bluetooth 2.0 profiles, kill Widcomm’s racketteering business, and open the door for compelling Bluetooth apps?