It depends where you are, and what you're looking for.
For me - many times, ebay is the only solution - as a buyer - simply as it's the only site with local sellers.
Craigslist for example - in the UK - specifically the middle of scotland.
Dundee + Edinburgh areas have around 6 items for sale per day.
(these cities are 40 miles or so from me)
Over all categories, ebay has well over a thousand within 25 miles.
Ebay is a functional monopoly in many cases.
For rarer stuff - for example - an x60s laptop with a broken screen - you simply won't find one for sale on craigslist - but one turns up every couple of weeks on ebay.
I wasn't commenting at all about the grammar - but the regrettable state of the patent system that seems - particularly in the US - to require a very low bar of novelty.
For example - a patent to keep dirt out of an enclosure containing a display - which had to not absorb accoustic waves.
The solution - a foam gasket with a film of plastic round it - is so mind numbingly obvious that anyone with a slight clue would try it about 4th or 5th thing from stuff lying around the workshop if they'd needed to solve the problem.
Umm - no. Low cost - 1mm accuracy - there are _lots_ of applications. Robot hoovers that don't random-walk - but navigate. Game industry - the wii would be lots better if it could have absolute position information.
Sports analysis - analyse your swing/... Robot lawnmowers, parking aids (to your garage), fork lift navigation,.....
That's not motion detection - if you're stationary. Also - before SA was turned off - if you were using standard GPSs, in order to get a average position within a meter, you needed to average for over a week. Unless you were - as you probably were - doing differential GPSs - which was using a seperate GPS at a known base-station to transmit errors due to selective availability, and correct the nearby surveying GPS.
The reason for GPSs getting lots better in the 90s was _not_ improvements in GPSs. It was the removing of 'Selective Availability' - which was an intentional up-to-300m (usually 100mish) random bias applied to the GPS signal.
In 2000 this was turned off.
Also - hardware improved so it could track multiple satellites at once - compared to one of the earliest units.
I first noticed this site a few days ago - linked from an ebay auction. http://cgi.ebay.ca/NIB--Asus-EeePC-1000HE-10"-160G%2F1GB-BLUE-EBONY-XP_W0QQitemZ290318394892QQcmdZViewItemQQimsxZ20090520?IMSfp=TL090520166001r25703 was one (ended) auction. - I was searching to find any 10" eeepcs with linux.
Umm - yes there is more to the internet than http. I specifically addressed this - gopher-with-pictures - in the post you commented on. My point is that everything was just right for the internet to take off.
There was widespread ownership of modems that could be used for it. The internet had been allowed to expand past academia alone, and given governmental funding. The users of online services hit their limits - especially in interconnects.
Take out any of these factors - and the internet doesn't grow, and possibly shrinks, or other corporate networks emerge.
Yes - because the internet grew fast enough, it won - email was probably the killer app - along with ftp. If it hadn't grown quite so fast, and aol and compuserv or some of the smaller players - had got effective interconnects working - at least for email, then the incentive for pick the internet instead of the AOLnet - or whatever would have been smaller.
If it was enough smaller, then the internet remains a small educational thing, until it - possibly - dies.
Six or twelve months delay in users being interested in the internet might have created a whole different future.
The users go where the content is.
If AOL and compuserv, and all the giants of old had gotten their own protocols/content built a _little_ faster than they had, then the exponential factor of the internets growth starts out much lower, or even negative.
In 1999, we could fairly easily have (IMO) seen a slower accepted gopher-with-pictures, and lack of Al Gores pushing of the funding of the eary internet having lead to the internet being a network used by some educational establishments, as a continuing research project, with buisnesses increasingly having presences on Aoluserv.
No it won't. The ST VS6555R0H9 for example. VGA, 4.5*4.5*2.5mm, I'd guess.25g tops.
There are slightly larger cameras going up to a couple of megapixels.
But you also need the decoder, the DSP, and the transmitter, and the battery.
Decoder, DSP, and micro I can directly address, as I have a board using a similar (but 2MP) cam about to go for production that is now about 4 grams, 5 including the microSD.
If I put maybe $400 into weight reduction, and build them by hand I can drop that to around 3.5 grams, including an 8GB microSD.
If you still manage to do all that, then what's the use of a low-res image from 400 km?
I'm tempted to drag out the famous quote 'What's the use of a newborn baby' - with regards to electricity, but that's way overstating the case.
At the moment, it's complex and hard for small organisations (say a medium sized electronics department in a university) to get anything launched.
I hope by vastly reducing the entry burden that more people will start innovating - where there is little at risk, there can be lots of risk-taking, much as the computer industry only exploded when computers got cheap enough that innovation in software got cheap.
Yes, you can beg for a ride-along, but this is not cheap in terms of effort you need to put in to it. (and that time costs money too).
Besides, rocketry is not a safe hobby when you deal with enough propellant to lift something to an LEO. When you try to do it on the cheap things only get scarier.
This is of course an issue.
However, constraining the rocket element to perhaps 10Kg, with a thrust profile meaning that it cannot hit the ground under power, selecting appropriate launch sites, and ensuring a low burnout weight will all greatly mitigate risks.
On the subject of mass budgets, which you raise.
Camera - 1g
Solar panel -4g (2*26cm^2 emcore) (3W peak, 0.5W average per orbit)
You're not going to get a nice big lens on there for a decent camera. But a cellphone cam might just work.
You can't get active control of your orientation. Though with the above camera, and a couple of magnetic field sensors, combined with software, you can tell which way you're pointed.
You won't be able to send back large amounts of data - but a small pic or two per orbit, commanded by the ground should barely be possible.
As to what can be done in something slightly bigger - the sky is the limit. (or not).
I note that people have made complete flying R/C planed under half a gram.
If these people had relatively cheap access to orbit, what might happen?
Firstly, you can't buy a kilo to orbit. You simply can't.
You may be able to beg a ride-along if you have the right political connections, but otherwise it's impossible.
Secondly, it's unlikely that if 20g to orbit is $2000, 200g to orbit will be $20000.
Thirdly, and perhaps most importantly.
20g to orbit can't do much. You can put a bad camera, a radio, a solar panel, and a magnetometer on it, and maybe if you push the envelope really hard a 3-axis gyro. (to calculate your orbit)
200g however, even if it was $10000 per flight is in the realm where universities with modest physics, aerospace, or electronics facilities might consider it interesting to put up a small test sat.
Your cellphone weighs under 200g, even if it has GPS, GSM, accellerometers, wifi, camera,...
With 200g in a small satellite, you've got a good shot at a reasonable camera, stabilisation using the earths magnetic field, GPS, a much better radio, solar panels, batteries to keep it alive during dark.
It's even reasonable that you could have a small part of it - say 50g - as a single-shot rocket able to optimise the trajectory.
The big difference is that Apple is probably going to try to patch all the holes as fast as they can.
The whole idea behind OpenMoko is that there is no need to use holes to get in. You are encouraged to get in, and to contribute if you can.
There is no OpenMoko nice toolkit at the moment because nobody has written one.
The manufacturer of the Neo1973 - the phone OM first runs on - has already provided drivers for every part of the phone, and you can run whatever application you like on it, once you compile it.
At the moment it is not a phone.
It's a slow x-windows desktop in a small case, with some rudimentary phone software on it.
Development works pretty much like any other x-windows program.
As to why FIC - the company behind it - wants to do this. At the moment they are a subcontractor, for example low-bidding on Nokias latest offering, and making it for them for a small profit. If they can sell direct to consumers, then their profit goes way up. Even if other companies start making phones.
Basically not.
If you hold the phone in the same orientation, with best-in-class accellerometers, you pretty much get a centimeter per second per second of error.
After an hour, that's 70MPH, and around 50 miles out.
It's useful for periods well under a minute, when the orientation is known.
If the orientation is unknown, then it's basically useless for navigation - even if you have two of them in the phone - for periods over 5 seconds.
http://wiki.openmoko.org/wiki/Accelerometer_Fundam entals - though it's somewhat slow at the moment.
Guess why:)
However, the ISP is _really_ screwed by their upstream. If they want to sell generic ISP service to anyone in the UK, then they pretty much have only one choice - pay BT for a 'central pipe' - this is a virtual pipe that goes from the linecard at your local exchange, to the ISPs NOC.
For this 622mbit/s pipe, the ISP is charged the very scary number of 150K pounds per quarter.
This is about 70 megabytes/second, once all the layers (ATM, IP encapsulation, maximum pipe loading) is taken into account.
Per month, this comes out to almost exactly 200000 gigabytes.
That is - assuming traffic patterns were ideal, it costs well over 50p per gigabyte to get that traffic back to the ISP.
Letting users max out 8mbit lines will cost you over 3000 pounds per month in bandwidth bills alone _PER USER_.
Assuming half revenue they get goes to bandwidth providers, half on staff, and half in 'profit' - that's around 3000/6 - 500 'regular users' paying the bills for one heavy user, before you start seeing _any_ possible profit.
Of course, most users will not do this.
But, even 30 gigabytes per month means that the ISP is certainly subsidising you, unless you're paying well above the normal market rates, or are in an area dense enough that it pays the ISP to install equipment in each exchange.
Completely open-source, though not ready for end-users at the moment. GPS (the GPS bit needs reverse engineered to remove a tiny bit of closed code), quadband GSM, nice dispay.
MP3 player, phone, trip computer, guitar tuner, games platform,...
You're only limted by your imagination. (and the hardware).
At the moment, on the Neo1973 - which is the phone that FIC is releasing first, you talk to the GSM modem via AT commands.
The dialer app is at the moment broken, and you use minicom or something:)
The only closed source bit of code that will ship with the phone is the code that takes the output from the very dumb GPS hardware, and 'cooks' it into an actual position. And there are moves to - when a working version of this is shipped, reverse engineer it, and make it open source too.
You can run _any_ 'normal' linux app on it, with the obvious limitations (no keyboard unless you've bluetooth, 2.8" display, touchscreen, one uncommitted button).
You can even put GCC, and a full normal toolchain on the microSD, and do native development work, if you really want to.
The problem is that - at the moment - it's not usefully open source.
They've open sourced a few drivers, which are admittedly handy, and may be useful for other stuff, but at the moment, most of the stuff they've open sourced is that that they are required to do, as they have used it in their product, and gained by not having to write their own kernel.
To be truly 'open source' in a meaningful way, you should at leat be able to add your own applications to the device, and ideally be able to completely rip out all the existing user interface stuff, and only use the camera, touchscreen, phone,... drivers to do whatever you want.
There are all sorts of possible nifty applications that you could do with this device, that are not available in the default application.
(Based on some assumptions about the hardware and installed software - I can't find a proper manual)
Time lapse photography.
Use any bluetooth keyboard to enter SMS/email
Connect to bluetooth GPS to do routing.
Do VOIP.
One-time-pad encrypted voice - 1Gb of flash is quite enough to store several one-time pads lasting hours each for talking to different people.
Photo, or voice-print authorisation to use the phone.
Email/... the real user if phone thinks it's stolen, if it is, turn on eavesdropping mode.
Print pictures to any bluetooth enabled printer.
Use any bluetooth mouse to 'paint' on pictures you've taken.
Notify the user if a radio program they are interested in is about to start.
Connect to the internet via bluetooth if it's available locally, rather than going over the phone network.
Offer internet connectivity over the phone network via bluetooth, for a small amount paypalled.
Speech synthesiser to read ebooks to you over bluetooth,.........
There is simply so much that could be done with the hardware, if it was properly open sourced.
Much of this sort of stuff is not put in by motorola et al, because the networks don't want it.
There are fundamental problems with self-managed 'grid' networks.
Basically, if you're all using the same frequency space (802.11*), then the overall random noise from the distant nodes - those far enough away that collision avoidance can't work, because you can't recieve their distinct signal, overrides local ones.
An example - consider a plane with a distribution of transmitters. Inside a certain radius, you can beat this a bit by doing the collision avoidance thing, but as you go outwards the signal from the nodes drops off by 1/r^2, but r more nodes appear, meaning that the contribution of each radius from the transmitter is not 1/r^2, but 1/r. Add all these up, and it sums to infinite noise
This can get better if the terrain, or atmosphere absorbs the frequency in question.
One way to stop this happening is to have several non-interfering networks overlaid.
For example, a wi-max network that carries 'long-haul' traffic off the 802.11* network.
Another problem with the 'grid' topology is that if the hops are 100m across, and you want to get to a major node that connects to the rest of the network 1Km away, the nodes that are right next to the big access point are each carrying the traffic of dozens of nearby nodes, for which they are the best route to the net.
Not to mention that it's not going to be this nice, because geography means that in nearly all areas you're not going to get a nice spread of traffic between nodes.
This is not a routing efficiency problem. It's a fundamental problem about the number of hops you need to connect through to get to the 'proper' internet. If the average connection takes 20 hops to get to the net, then at best, you're looking at each node having 5% of it's sticker bandwidth.
Interesting table below - showing maximum time at given G levels. Interestingly, from the second line and below, you're at about orbital speed at the end of the accelleration, if it's linear.
Summary of Results: (Data primarily from: Bioastronautics Data Book,
second edition, 1973, NASA)
I - Sustained Acceleration
CAPTION: Time vs. Tolerance : G-force limits for reclining subjects
Slashdot Mirror
So 3G phones magically use some other means than radio that's not shared?
It depends where you are, and what you're looking for.
For me - many times, ebay is the only solution - as a buyer - simply as it's the only site with local sellers.
Craigslist for example - in the UK - specifically the middle of scotland.
Dundee + Edinburgh areas have around 6 items for sale per day.
(these cities are 40 miles or so from me)
Over all categories, ebay has well over a thousand within 25 miles.
Ebay is a functional monopoly in many cases.
For rarer stuff - for example - an x60s laptop with a broken screen - you simply won't find one for sale on craigslist - but one turns up every couple of weeks on ebay.
The panel the multimeter is resting on is a conventional silicon solar panel.
It's the right colour - a flat black.
It's the conventional shape.
It's almost exactly on the 16% or so efficiency - based on the 18W output quoted and its apparant size - of a normal silicon panel.
9V is about the right voltage for that size of panel.
The 'hair' pictured looks a hell of a lot like enamelled copper wire.
If it quacks like a duck, walks like a duck, and makes love like a duck, it's a duck.
I wasn't commenting at all about the grammar - but the regrettable state of the patent system that seems - particularly in the US - to require a very low bar of novelty.
For example - a patent to keep dirt out of an enclosure containing a display - which had to not absorb accoustic waves.
The solution - a foam gasket with a film of plastic round it - is so mind numbingly obvious that anyone with a slight clue would try it about 4th or 5th thing from stuff lying around the workshop if they'd needed to solve the problem.
Non-obvious is a high bar nowadays?
I would comment further, but I just soiled myself laughing, and I need to shower.
Umm - no.
Low cost - 1mm accuracy - there are _lots_ of applications.
Robot hoovers that don't random-walk - but navigate.
Game industry - the wii would be lots better if it could have absolute
position information.
Sports analysis - analyse your swing/... .....
Robot lawnmowers, parking aids (to your garage), fork lift navigation,
No, it's not - that's not the point.
The point is that _only_ google gets this right. Nobody else.
You trust google now.
Are you sure you will trust them in 30 years?
Why shouldn't others be able to scan books and make them available in the same way google is?
That's not motion detection - if you're stationary.
Also - before SA was turned off - if you were using standard GPSs, in order to get a average position within a meter, you needed to average for over a week.
Unless you were - as you probably were - doing differential GPSs - which was using a seperate GPS at a known base-station to transmit errors due to selective availability, and correct the nearby surveying GPS.
The reason for GPSs getting lots better in the 90s was _not_ improvements in GPSs.
It was the removing of 'Selective Availability' - which was an intentional up-to-300m (usually 100mish) random bias applied to the GPS signal.
In 2000 this was turned off.
Also - hardware improved so it could track multiple satellites at once - compared to one of the earliest units.
I first noticed this site a few days ago - linked from an ebay auction.
http://cgi.ebay.ca/NIB--Asus-EeePC-1000HE-10"-160G%2F1GB-BLUE-EBONY-XP_W0QQitemZ290318394892QQcmdZViewItemQQimsxZ20090520?IMSfp=TL090520166001r25703 was one
(ended) auction. - I was searching to find any 10" eeepcs with linux.
Umm - yes there is more to the internet than http.
I specifically addressed this - gopher-with-pictures - in the post you commented on.
My point is that everything was just right for the internet to take off.
There was widespread ownership of modems that could be used for it.
The internet had been allowed to expand past academia alone, and given governmental funding.
The users of online services hit their limits - especially in interconnects.
Take out any of these factors - and the internet doesn't grow, and possibly shrinks, or other corporate networks emerge.
Yes - because the internet grew fast enough, it won - email was probably the killer app - along with ftp.
If it hadn't grown quite so fast, and aol and compuserv or some of the smaller players - had got effective interconnects working - at least for email, then the incentive for pick the internet instead of the AOLnet - or whatever would have been smaller.
If it was enough smaller, then the internet remains a small educational thing, until it - possibly - dies.
The internet winning was not a certainty.
Six or twelve months delay in users being interested in the internet might have created a whole different future.
The users go where the content is.
If AOL and compuserv, and all the giants of old had gotten their own protocols/content built a _little_ faster than they had, then the exponential factor of the internets growth starts out much lower, or even negative.
In 1999, we could fairly easily have (IMO) seen a slower accepted gopher-with-pictures, and lack of Al Gores pushing of the funding of the eary internet having lead to the internet being a network used by some educational establishments, as a continuing research project, with buisnesses increasingly having presences on Aoluserv.
Look at for example what happened in france, with minitel - http://en.wikipedia.org/wiki/Minitel - it held its own well into the late 90s.
The web - helped the internet to win - although admittedly it came rather late in the game, and was among the final nails in the proprietory networks.
But a cellphone cam might just work.
A cell phone CCD will be about 20 grams.
No it won't. .25g tops.
The ST VS6555R0H9 for example. VGA, 4.5*4.5*2.5mm, I'd guess
There are slightly larger cameras going up to a couple of megapixels.
But you also need the decoder, the DSP, and the transmitter, and the battery.
Decoder, DSP, and micro I can directly address, as I have a board using a similar (but 2MP) cam about to go for production that is now about 4 grams, 5 including the microSD.
If I put maybe $400 into weight reduction, and build them by hand I can drop that to around 3.5 grams, including an 8GB microSD.
If you still manage to do all that, then what's the use of a low-res image from 400 km?
I'm tempted to drag out the famous quote 'What's the use of a newborn baby' - with regards to electricity, but that's way overstating the case.
At the moment, it's complex and hard for small organisations (say a medium sized electronics department in a university) to get anything launched.
I hope by vastly reducing the entry burden that more people will start innovating - where there is little at risk, there can be lots of risk-taking, much as the computer industry only exploded when computers got cheap enough that innovation in software got cheap.
Yes, you can beg for a ride-along, but this is not cheap in terms of effort you need to put in to it. (and that time costs money too).
Besides, rocketry is not a safe hobby when you deal with enough propellant to lift something to an LEO. When you try to do it on the cheap things only get scarier.
This is of course an issue.
However, constraining the rocket element to perhaps 10Kg, with a thrust profile meaning that it cannot hit the ground under power, selecting appropriate launch sites, and ensuring a low burnout weight will all greatly mitigate risks.
On the subject of mass budgets, which you raise.
What can be done in 20g?
What can't?
There are obvious limits.
You're not going to get a nice big lens on there for a decent camera. But a cellphone cam might just work.
You can't get active control of your orientation. Though with the above camera, and a couple of magnetic field sensors, combined with software, you can tell which way you're pointed.
You won't be able to send back large amounts of data - but a small pic or two per orbit, commanded by the ground should barely be possible.
As to what can be done in something slightly bigger - the sky is the limit. (or not).
I note that people have made complete flying R/C planed under half a gram.
If these people had relatively cheap access to orbit, what might happen?
Cost per kilo is somewhat missing the point.
Firstly, you can't buy a kilo to orbit. You simply can't.
You may be able to beg a ride-along if you have the right political connections, but otherwise it's impossible.
Secondly, it's unlikely that if 20g to orbit is $2000, 200g to orbit will be $20000.
Thirdly, and perhaps most importantly.
20g to orbit can't do much. You can put a bad camera, a radio, a solar panel, and a magnetometer on it, and maybe if you push the envelope really hard a 3-axis gyro. (to calculate your orbit)
200g however, even if it was $10000 per flight is in the realm where universities with modest physics, aerospace, or electronics facilities might consider it interesting to put up a small test sat.
Your cellphone weighs under 200g, even if it has GPS, GSM, accellerometers, wifi, camera, ...
With 200g in a small satellite, you've got a good shot at a reasonable camera, stabilisation using the earths magnetic field, GPS, a much better radio, solar panels, batteries to keep it alive during dark.
It's even reasonable that you could have a small part of it - say 50g - as a single-shot rocket able to optimise the trajectory.
I note that http://www.rcgroups.com/forums/showthread.php?t=814157 there are amateur build fully remote controlled planes at under half a gram.
She was introduced to him in russia by a marriage agency, for the purposes of getting out of russia.
The big difference is that Apple is probably going to try to patch all the holes as fast as they can.
The whole idea behind OpenMoko is that there is no need to use holes to get in.
You are encouraged to get in, and to contribute if you can.
There is no OpenMoko nice toolkit at the moment because nobody has written one.
The manufacturer of the Neo1973 - the phone OM first runs on - has already provided drivers for every part of the phone, and you can run whatever application you like on it, once you compile it.
At the moment it is not a phone.
It's a slow x-windows desktop in a small case, with some rudimentary phone software on it.
Development works pretty much like any other x-windows program.
As to why FIC - the company behind it - wants to do this.
At the moment they are a subcontractor, for example low-bidding on Nokias latest offering, and making it for them for a small profit.
If they can sell direct to consumers, then their profit goes way up.
Even if other companies start making phones.
Basically not. If you hold the phone in the same orientation, with best-in-class accellerometers, you pretty much get a centimeter per second per second of error. After an hour, that's 70MPH, and around 50 miles out. It's useful for periods well under a minute, when the orientation is known. If the orientation is unknown, then it's basically useless for navigation - even if you have two of them in the phone - for periods over 5 seconds. http://wiki.openmoko.org/wiki/Accelerometer_Fundam entals - though it's somewhat slow at the moment.
Guess why :)
However, the ISP is _really_ screwed by their upstream. If they want to sell generic ISP service to anyone in the UK, then they pretty much have only one choice - pay BT for a 'central pipe' - this is a virtual pipe that goes from the linecard at your local exchange, to the ISPs NOC.
For this 622mbit/s pipe, the ISP is charged the very scary number of 150K pounds per quarter.
This is about 70 megabytes/second, once all the layers (ATM, IP encapsulation, maximum pipe loading) is taken into account.
Per month, this comes out to almost exactly 200000 gigabytes.
That is - assuming traffic patterns were ideal, it costs well over 50p per gigabyte to get that traffic back to the ISP. Letting users max out 8mbit lines will cost you over 3000 pounds per month in bandwidth bills alone _PER USER_.
Assuming half revenue they get goes to bandwidth providers, half on staff, and half in 'profit' - that's around 3000/6 - 500 'regular users' paying the bills for one heavy user, before you start seeing _any_ possible profit.
Of course, most users will not do this.
But, even 30 gigabytes per month means that the ISP is certainly subsidising you, unless you're paying well above the normal market rates, or are in an area dense enough that it pays the ISP to install equipment in each exchange.
I've been reluctant to buy a new MP3 player, because there was no supported linux (native) one.
Now, I'm getting a http://wiki.openmoko.org/ Neo1973 phone, come late march, when they go on sale.
Completely open-source, though not ready for end-users at the moment. GPS (the GPS bit needs reverse engineered to remove a tiny bit of closed code), quadband GSM, nice dispay.
MP3 player, phone, trip computer, guitar tuner, games platform, ...
You're only limted by your imagination. (and the hardware).
http://rapidshare.com/files/18781887/rect.avi A one hour talk by one of the creators of the OS, OpenMoko.
The reason a phone maker - FIC - is interested in this is simple.
At the moment, they are one of several firms asked to bid to make a new phone from Nokia et al.
This makes little money.
If they can get into selling phones directly to end users, then they make _far_ more profit. Even if other people start releasing OpenMoko phones.
OpenMoko is the OS, there may be many phones.
At the moment, on the Neo1973 - which is the phone that FIC is releasing first, you talk to the GSM modem via AT commands.
The dialer app is at the moment broken, and you use minicom or something :)
The only closed source bit of code that will ship with the phone is the code that takes the output from the very dumb GPS hardware, and 'cooks' it into an actual position. And there are moves to - when a working version of this is shipped, reverse engineer it, and make it open source too.
You can run _any_ 'normal' linux app on it, with the obvious limitations (no keyboard unless you've bluetooth, 2.8" display, touchscreen, one uncommitted button).
You can even put GCC, and a full normal toolchain on the microSD, and do native development work, if you really want to.
(think of a Pentium 100 laptop sort of speed)
http://wiki.openmoko.org/ - the wiki. http://rapidshare.com/files/18781887/rect.avi a 1 hour talk (60M) on OpenMoko, by one of the instigators of the OpenMoko project.
They've open sourced a few drivers, which are admittedly handy, and may be useful for other stuff, but at the moment, most of the stuff they've open sourced is that that they are required to do, as they have used it in their product, and gained by not having to write their own kernel.
To be truly 'open source' in a meaningful way, you should at leat be able to add your own applications to the device, and ideally be able to completely rip out all the existing user interface stuff, and only use the camera, touchscreen, phone, ... drivers to do whatever you want.
There are all sorts of possible nifty applications that you could do with this device, that are not available in the default application.
(Based on some assumptions about the hardware and installed software - I can't find a proper manual)
Time lapse photography.
Use any bluetooth keyboard to enter SMS/email
Connect to bluetooth GPS to do routing.
Do VOIP.
One-time-pad encrypted voice - 1Gb of flash is quite enough to store several one-time pads lasting hours each for talking to different people.
Photo, or voice-print authorisation to use the phone.
Email/... the real user if phone thinks it's stolen, if it is, turn on eavesdropping mode.
Print pictures to any bluetooth enabled printer.
Use any bluetooth mouse to 'paint' on pictures you've taken.
Notify the user if a radio program they are interested in is about to start.
Connect to the internet via bluetooth if it's available locally, rather than going over the phone network.
Offer internet connectivity over the phone network via bluetooth, for a small amount paypalled.
Speech synthesiser to read ebooks to you over bluetooth, .........
There is simply so much that could be done with the hardware, if it was properly open sourced.
Much of this sort of stuff is not put in by motorola et al, because the networks don't want it.
Basically, if you're all using the same frequency space (802.11*), then the overall random noise from the distant nodes - those far enough away that collision avoidance can't work, because you can't recieve their distinct signal, overrides local ones.
An example - consider a plane with a distribution of transmitters. Inside a certain radius, you can beat this a bit by doing the collision avoidance thing, but as you go outwards the signal from the nodes drops off by 1/r^2, but r more nodes appear, meaning that the contribution of each radius from the transmitter is not 1/r^2, but 1/r. Add all these up, and it sums to infinite noise
This can get better if the terrain, or atmosphere absorbs the frequency in question.
One way to stop this happening is to have several non-interfering networks overlaid.
For example, a wi-max network that carries 'long-haul' traffic off the 802.11* network.
Another problem with the 'grid' topology is that if the hops are 100m across, and you want to get to a major node that connects to the rest of the network 1Km away, the nodes that are right next to the big access point are each carrying the traffic of dozens of nearby nodes, for which they are the best route to the net.
Not to mention that it's not going to be this nice, because geography means that in nearly all areas you're not going to get a nice spread of traffic between nodes.
This is not a routing efficiency problem. It's a fundamental problem about the number of hops you need to connect through to get to the 'proper' internet. If the average connection takes 20 hops to get to the net, then at best, you're looking at each node having 5% of it's sticker bandwidth.
Interesting table below - showing maximum time at given G levels.
Interestingly, from the second line and below, you're at about orbital speed
at the end of the accelleration, if it's linear.
Summary of Results: (Data primarily from: Bioastronautics Data Book,
second edition, 1973, NASA)
I - Sustained Acceleration
CAPTION: Time vs. Tolerance : G-force limits for reclining subjects
Time (min) Accel Coach (Gs) Water Immersed (Gs)
0.5 23 28
1.0 17 22
2.0 12 17
3.0 8 14
4.0 7 12
5.0 6 9
6.0 5 8
7.0 4 7
8.0 4 6
9.0 3 6
10.0 3 6