About the size of an RFID. The temperature sensor was an off the shelf MEMS in an IC package, there was the main PIC as a processor and some other odds and ends to link it to the bus. We weren't using Ziggbee, but that is a single package now so I'm thinking the whole thing could be a lot smaller than a cell phone. The battery was just a watch battery (don't know what they are called, the little circular ones). The power draw from the processor, sensor and bus was practically zero. Essentially it is a question of how big a battery is needed to fire up the wireless once a minute and squirt out a packet.

The link is still broken so here are some semi-random thoughts on what kind of remote control I was looking for a few days ago. I want a reasonably accurate temperature sensor, not +- 1 or even 2 degrees C. I want +-0.1 degrees with a update frequency of about once a minute. Wireless with a 50-100m range to a base-station that can then log the data. I don't want some kind of "cloud service" where the base-station uploads my data to the maker's server. I don't want a closed proprietary interface. I want the battery in the sensor to last at least a year, and the sensors should be dirt cheap (less than $20) so that I can put one in every room in the house and map heat flow through the rooms.

Why does this not exist? We were building these kinds of devices in a lab ten years ago, I assumed they would have been commercialised by now. Instead the home automation products that I found were outrageously expensive and limited (proprietary interface to a private server, all access via web). There are no technical limitations, and the parts are cheap enough to make a decent margin on $20 for a PIC microchip, a Zigbee controller and an IC for temperature. It does seem weird that the market that exists for home automation is so distorted.

Are you assuming that Schrodinger never watched his collection?

Making the install size of the system smaller enables new applications and ways of using the system. At the moment an operating system is a giant monolithic install tied to a signle machine. I would prefer a single install shared across all of my devices. To make this work the OS would need to be synchronised / reconciled across a network connection on demand. Like a cloud OS but not running on public hardware - just migrating between devices that I own. As size as a direct impact on performance (time to start up, to reconcile) the install size could never be small enough.

There are smaller devices than you acknowledge. Memory and storage are the dominant costs in those devices and reducing the need for both enables cheaper classes of device. The Pi is an interesting machine for $25, but what happens when a desktop-capable machine costs $5, or $1?

Lastly, flash is nice but it still is not as fast as DRAM. Having an entire OS install cached in memory has advantages for performance.

When you have a filesystem that understands hard links, deduplication is still required to find files that have the same content and link them together. You are possibly thinking of a filesystem that hashes contents to decide on storage locations.

You keep talking about a large asteroid that is definitely on a collision course with earth. This is an error. There have been large asteroids that been on collision courses before. There may be more, but it is not certain. Because it is not 100% probable that an unknown large asteroid is currently on a collision course with earth none of your argument is valid, and your arguments about increasing probabilities are incorrect.

Your argument would be valid (but pointless) if you said: "if there is at least one more asteroid on a collision course with earth then the probabillty of impact on a particular day increases with each day it does not strike". Of all the asteroids that strike earth of a given size, one of them is the last.

If you think fuzzy is bad look at anything called quazi-something in maths. It may as well read not-actually-anything-at-all-like-something. That one is my pet peeve

When people write a paper for publication they have to differentiate their approach from previous approaches. You seem to have latched onto deep as an imprecise description of the number of layers. It is not. It is an accurate distinction in comparison to previous approaches. Because previous approaches were limited to about (not exactly) two layers it makes the definition of the label a little fuzzy, but the partition into shallow / deep approaches is crisp.

The problem comes when you try larger inputs. Regardless of constant factors if you are playing with O(2^n) algorithms then n will not increase above about 30. If you start looking at really weird stuff (optimal circuit design and layout) then the core algorithms are O(2^2^n) and then if you are really lucky n will reach 5. Back in the 80s it only went to 4, buts thats Moore's law for you.

When you refer to this hypothetical steambox as open, what makes you assume that it will be? If it is custom hardware there is no reason to assume that it will be any more open than competing consoles. Linux on the inside doesn't guarantee root access, and if they drop bios for a signed bootloader it would be difficult to get into. The DRM in the system will be derived from steam, so I would assume that online activation will be mandatory. The piracy rate could actually be lower than the currently existing consoles as none of them enforce mandatory online activiation. It could also completely kill the second-hand game market for this device, so the economics of the market would quite new. As you wrote ealier, it would be up to Value to prove that gamers would buy into this market, although I assume the studios would love it.

Was about to post this. Also steal a part of ringbring's business model and pick up other supplies on route: popcorn, chocolate, soda, pizza etc.

Probably more showers... But they are less fun.

(that I'm too lazy to read the article for) is does this mean an end to passwords, and the beginning of proper identity management in the browser. I would love to authenticate myself once to the browser, and then have it negotiate identities with each website.

We have motorised desks at work. There is a pair of push switches on the front and they raise and lower maybe 1.5m (3-4ft). As long as you have fairly long cables on everything these make the transition very easy. It does make quite a difference as you can stretch your legs for an hour or two and then switch back. Oddly enough I've ended up in a weird slouched position with the chair reclined and the desk close to the floor, meh, the broken ergonomics are probably more to do with me than the office.

Fair enough, I think that I skipped over your penultimate paragraph. I'm not the poster that you were replying to though...