4 ms·

I fully expect these to be commodity parts made by the likes of Dell or HP inside of 8-12 years. Setting aside the fact that neither Dell nor HP actually make

I fully expect these to be commodity parts made by the likes of Dell or HP inside of 8-12 years. Setting aside the fact that neither Dell nor HP actually make any of the parts inside their PCs now, I think there are too many separate issues that you're bringing up here. Clarifying, 1. Will netbooks (you mentioned the OQO) be built on modular, user-swappable hardware components. 2. Will smartphones be built on modular, user-swappable hardware components (you said, " 802.11 if it is cheap and efficient enough, but provide a slot for drop-in 4g, bluetooth, DRM ") 3. Will smartphones be build according to a fixed hardware/software HAL that is well specified enough that no OS or driver customization is needed and user-installable and substitutable OS's are possible. #1 is possible IMHO. A few years ago ASUS was selling a customizable laptop platform that small computer stores could build and sell as their own brand. It was big and bulky and went away quickly, which is always going to be a problem with modular hardware, so I don't see this as ever being very popular. But someone might try it again with netbooks. #2 is just never going to happen. Forget about it. if you provide a slot for drop-in 4g, bluetooth, DRM, someone else is going to build a competing smartphone with all of that built in but without the bulky connector and housing. It'll be smaller and sexier than your monster-smartphone, and it'll probably have better battery life with the extra internal room for a larger battery. Modularity is irrelevant when you can just include everything that's cutting-edge at the moment and you'll just get a new smartphone in a couple of years under a new contract anyways. #3 might happen, but I still doubt it. It goes back to what I said about constraints and room for innovation. When something big needs to change with the PC HAL (like USB3), Intel, Microsoft, Dell, and a bunch of others have to get together and formulate something like XHCI to provide a common hardware interface - otherwise every USB3 chip would require a separate loadable driver. In the embedded market and with smartphones, there's no constraint on the hardware interface - you just throw the chip in and then add a kernel driver for it. Time saved, competitive advantage gained. How much longer would it have taken for NFC to be included in smartphones if Samsung, HTC, Google, Apple, Microsoft, Nokia, and others had to get together and hash out a common hardware interface for NFC chips the way PC manufacturers do with stuff like UHCI/EHCI/AHCI/etc? A lot longer. And given that no one expects user-installable and substitutable OSs on smartphones, there's no incentive now or in the future to provide one. And given that a smartphone platform that provides a common HAL is going to be at a competitive disadvantage wrt smartphones with custom hardware, there's no thin end to this wedge.

rch 16y ago
First, thanks for the nice response. "Setting aside the fact that neither Dell nor HP actually make any of the parts inside their PCs now" I was referring to the whole device as a 'part', and don't really expect Dell or HP take over from Broadcom, TI, and so on. Anyone that sells a tablet could sell this. 1: Hopefully not. Ideally, hardware should perform well for the lifetime of the device, without having to swap out any of the internals. I gave up screwdrivers a few years ago. 2: I don't think I said smartphones... but you could use the slot to turn whatever it is into a phone, if you really need it. Personally, I will use something akin to Google voice or Skype over 802.11, but mobile data connections are fine. It needn't be bulky either -- the smaller express card form factor was nice, for instance. 3: Can you run your OS on Xen, etc.? Then everything will be fine. A good prototype device would run an AMI directly from your AWS account... a killer device would be able to expose exotic integrated hardware features when available, but it would be up to an OS to deal with it. Now, that's obviously more difficult, but not as difficult it was ten years ago. Point taken on the NFC example. In my world though, people only need to agree at the VM level.

rch 16y ago

First, thanks for the nice response. "Setting aside the fact that neither Dell nor HP actually make any of the parts inside their PCs now" I was referring to the whole device as a 'part', and don't really expect Dell or HP take over from Broadcom, TI, and so on. Anyone that sells a tablet could sell this. 1: Hopefully not. Ideally, hardware should perform well for the lifetime of the device, without having to swap out any of the internals. I gave up screwdrivers a few years ago. 2: I don't think I said smartphones... but you could use the slot to turn whatever it is into a phone, if you really need it. Personally, I will use something akin to Google voice or Skype over 802.11, but mobile data connections are fine. It needn't be bulky either -- the smaller express card form factor was nice, for instance. 3: Can you run your OS on Xen, etc.? Then everything will be fine. A good prototype device would run an AMI directly from your AWS account... a killer device would be able to expose exotic integrated hardware features when available, but it would be up to an OS to deal with it. Now, that's obviously more difficult, but not as difficult it was ten years ago. Point taken on the NFC example. In my world though, people only need to agree at the VM level.