It shouldn't be necessary to say this, but rsh is an enormous security hole! It authenticates using IP addresses only. Use ssh instead, it has the same functionality.
Great idea about the spotlight! This reminds me, I was once on a training course where the instructor had a great way of dealing with people who took mobile phone calls during sessions: he said "We'll wait for you" while they took the call. Only happened once due to the embarrassment of keeping your peers waiting...
Using 3x10^8 m/s for speed of light, I make it 0.1 ms = 100 usec one-way to the balloon, i.e. 0.2 ms for ground-to-ground via the balloon. So the wireless part will be fine as far as Quake is concerned - it's just the other parts of the trip that may add latency.
In practice, latency may be much worse than this - look at GPRS, where latencies are many hundreds of milliseconds (often half a second) even though the wireless distances are much shorter. Wireless links use a lot of extra error correction and in the case of GSM/GPRS, packets are sometimes split into much smaller frames that are then interleaved with other frames - this is because GSM is voice-centric and this works well for voice. Not sure if GPRS has a better coding that avoids this, but if this balloon system is designed for data it should avoid some of the overhead.
GPRS will be everywhere that GSM exists today (i.e. available in almost every country), a fairly slow but always-available packet-based transport - it is sometimes unreliable at present at least in the UK, but it generally works OK and is fast enough for short emails and suitably small web pages. Great for killing time if you are waiting to meet someone and have exhausted your AvantGo offline reading.
GPRS will also be upgraded to EDGE in many countries, tripling capacity and improving performance a lot. W-CDMA/UMTS (i.e. true 3G for GSM operators) is much more expensive to roll out but will provide more flexibility and performance - the real issue is whether 3G will make enough money to pay back its investment.
802.11b/Wi-Fi is not really an alternative to GPRS - there is no way there will be hot spots covering the whole of a large city, which is what GPRS does today. Wi-Fi is great for use in hotspots and within businesses - the ideal is a phone or PDA that has Wi-Fi, GPRS and perhaps 3G, switching between them based on availability and cost. Nokia has already released a PC card that does Wi-Fi and GPRS, and one trend is to put GSM SIM cards (smartcards) into Wi-Fi PC Cards, enabling you to automatically bill your Wi-Fi usage to your GSM phone account.
Finally, this camera chip is of course not dependent on GPRS, it will work fine with any wireless technology including CDMA2000. Most of what I've said above applies to CDMA, with CDMA2000 1x instead of GPRS and CDMA2000 1xEV-DO instead of W-CDMA - the only difference really is that CDMA operators have a much easier upgrade path, and don't need much spectrum, but they will probably remain in the minority world wide compared to the 70% market share of GSM globally.
I think picture messaging (part of MMS [Multimedia Messaging Service] in GPRS networks) will be huge, and this camera chip will help it take off. If you can send a picture to anyone with an MMS mobile phone or an email account, you can send postcards to friends and 'how do I fix this' messages to suitable experts, and get 'top 5 goals' messages, photos from Internet personal ads, etc... The more people have a camera built into their phone, the more they will use it (though probably never as much as plain text SMS).
MMS phones are already available in Europe (Ericsson T68i, with Nokia 7650 soon) where MMS is just starting - in Japan, J-Phone has had a huge success with picture messaging, known as Sha-Mail (over 4 million picture messaging handsets sold). Watch this space...
Even if you don't have a mobile phone, you'll be able to send email with picture/sound/video attachments to anyone with an MMS phone.
You are exaggerating - I live in London too, and the trains and tubes are not that bad (at least from where I live). Calling the trains 'a death trap' is ironic when you consider that railway transport is one of the safest ways to travel - far more people die every year on the roads, but this is virtually invisible since it happens in many small accidents, compared to the large and well publicised rail accidents.
The real solution is time shifting and working from home - I commute into work at 9 to 9.30 am and the tubes are nearly empty.
This only works in Windows where you have identical apps everywhere. Unix workstation setups can have all apps installed on servers but run on the clients, meaning that an engineer gets a unique set of apps when they log on to the same workstation just used by a manager.
You probably have to see this to realise how much better it is than Windows.
GPRS also allows for billing by bandwidth usage - the real innovation in NTT DoCoMo's i-mode was the business model (NTT only taking 9% of total revenues vs 50% in Europe), which stimulated a lot of new content providers. Official providers bill based on content, via NTT. (I think CDMA2000 1x will also bill by bandwidth used, but it depends on the operator).
Other innovations that helped i-mode include the end-to-end design (handset, protocols, infrastructure, billing, business model, content selection and marketing all from a single company) - like the Macintosh, the result is an easy to use and attractive product, although quite proprietary. Another key factor is that many Japanese commuters spend 2 to 4 hours per day on overground trains, making i-mode ideal for killing time. There is probably also some truth in the idea that PCs and fixed line Internet didn't take off so fast in Japan, so i-mode phones were filling a gap, and also fitted into smaller Japanese apartments.
Some of these factors can be replicated in the US and Europe - I'm interested in trying something that actually works consistently and has usable and interesting content (i.e. nothing like WAP...).
Yes, the US phone system is so great that I regularly get just 20 to 25 Kbps on a 56K modem there, whereas in the UK it's always over 40 Kbps. (Irony alert...)
The best thing about the US phone system is the flat-rate for local phone calls, but that's also why cellular operators have to charge for incoming phone calls to mobile phones in the US (otherwise they'd lose money on every call). And this incoming call charging is a key reason why US residents don't give out their cell phone numbers, reducing the overall size of the cellular market and thus mobile penetration. This is the real reason for low mobile phone usage, it has very little to do with the quality of land lines, only their cost vs cell phones.
Japan is the only country where I saw ISDN data sockets on payphones everywhere, even tiny ski resorts, and voice quality was fine when I was there. If you really want to talk crap land line systems, try India - the GSM coverage is not too great but at least it works better than the very noisy land lines.
What would be really useful is a Bluetooth-connected Springboard dock, i.e. you plug your Springboard into a separate gizmo and Bluetooth networks it to your new organiser (Treo, Palm, Sony, etc). A bit slower probably, but OK for many applications, just needs software support in PalmOS to extend the Springboard interface over Bluetooth.
The only flaw in this idea is that Handspring have for some reason not put Bluetooth in their devices, unlike most new mid-to-high end phones in Europe, but Sony and Palm do have Bluetooth. They also forgot GPRS for some reasons - no idea why, since GPRS is available in every major European country and is rolling out in the US.
I have a Visor Deluxe but I just bought a Palm m515 so I can use Bluetooth to network to my GPRS phone (Ericsson T68). At the rate wireless technology is evolving, I think separate PDA and phone is the way to go - this also makes it easy to network a laptop to the phone, and laptop to PDA for syncing, etc. You can also use Bluetooth for wireless headsets for extra geek points...
You can already sign routing updates with BGP (using MD5 authentication) and probably with other protocols as well. BGP is the most important one of course because it is used between ISPs.
The Swedish warship, Vasa, also failed due to unrealistic timescales and lack of requirements validation. Many of these technology failures are really process/project management failures, of course.
I saw the Vasa in its museum the other week in Stockholm - they retrieved the ship from the bottom of the harbour and it is now on display, with very interesting exhibits about how it was built. Worth a visit if you are ever in Stockholm.
See http://panipcase.homeip.net - already posted by someone else but buried within a thread - this is a site for several companies already being sued by PanIP. They are all small companies of course, but if they club together and get help from the EFF etc, they could probably beat PanIP and countersue. It would be well worth signing up there and commenting on their discussion forum.
Ultimately, the provider can always monitor how much bandwidth you are using by looking at its own routers - you can't spoof this. Search for 'Cisco NetFlow' for one example of how to do this.
By making it more expensive for them to detect cable modem uncapping, you are probably just going to encourage them to disconnect uncappers rather than just warning them.
GPRS, used by Voicestream, Cingular and AT&T Wireless, is a competing technology to CDMA2000 1x (i.e. what Verizon uses). I'm sure CDMA2000 services can be billed by the Kbyte, it's just that Verizon has chosen the older per-minute model. Every GPRS operator I know of (mainly in Europe, it's an upgrade of the global GSM mobile phone standard) uses per-Kbyte billing, to encourage people to stay on the network at all times. GPRS phones do have a 'connect' phase, but you can stay connected for days with no extra charges.
Anyway - the key point is that GPRS and CDMA2000 are both *technically* always-on, but may be implemented with a billing scheme that discourages always-on. Perhaps this is more common in the US, but IMO it's a big mistake. Once people are always-on, they can start doing ad-hoc messaging and web browsing whenever they feel like it, without the connection delay.
This comes down to the service creation APIs that live on servers in the 3G network - while 3G has a huge range of such APIs, and a lot of capabilities in this area, it's rather doubtful that these will be fully exposed to the customer. UMTS Release 5 (a later release of the main 3G standards) will be wholly based on SIP, but you'll still need to go through a SIP proxy).
However, at least with EPOC APIs for C++ and Java you'll be able to fully program whatever features are exposed from the network.
I live in London, and the air quality here is appalling compared to most US cities I've visited - over there, I can actually *smell* the car fumes are cleaner...
Have to agree about transport though - despite the hassles of public transport in London, it is still easy enough to get around without a car.
Thanks for the pointer - Dreamhost look really good, and I've been researching web hosts for some time. Particularly impressive that they pre-install many optional Perl modules, include SSH and 100 MB space in the $10/month account, and have a very good knowledge base.
They look particularly good for more complex Perl CGI systems, e.g. TWiki, a web collaboration system I'm working on (see http://twiki.org).
It's certainly trivial to spend lots of money on a startup, but it's not essential. The programmers can work from home to start with, the execs can do their own admin, and so on. It doesn't help that the IDC report came out, but IMO running out of money that early is just bad planning.
Local calls are only 'free' in the US if you make enough to cover the monthly local calling fee (at whatever rate would be charged). I don't make a lot of local calls in the UK since I got ADSL, and just use my mobile for most social/business calls, so the US 'free local calls' scheme would actually cost me money.
The rest of the world is actually *bigger* than the US (shock!) and GSM works just fine there...
This is ridiculous - operators are NOT worried about this sort of thing. Have you thought about how hard it would be to modify a modern phone, with its incredibly tight packing of circuitry and large-scale integration, to be a scanner?
There are plenty of operators making money in Europe without needing to lock phones (though some of them still do).
Operators simply want to make it harder for people to switch, even though the subscriber is already tied in by a contract (meaning that the operator's phone subsidy is not a reason for phone locking - the contract ensures they will make enough money to cover the subsidy).
BT Cellnet (O2) and Vodafone have never locked phones in the UK, and nor do Virgin. Orange and one2one (now T-Mobile) do lock their phones. However, it doesn't cost much to unlock the phones, as long as your initial contract (typically 12 months) is finished.
Phone locking is fairly pointless IMO, and should be stopped - the fact that Vodafone, who are a very profitable wireless operator, don't do it shows that it's not necessary. After all, people are already locked in by the contract, so all you are doing is stiffing them for an extra 20 UKP or whatever to unlock.
Slashdot Mirror
It shouldn't be necessary to say this, but rsh is an enormous security hole! It authenticates using IP addresses only. Use ssh instead, it has the same functionality.
Great idea about the spotlight! This reminds me, I was once on a training course where the instructor had a great way of dealing with people who took mobile phone calls during sessions: he said "We'll wait for you" while they took the call. Only happened once due to the embarrassment of keeping your peers waiting...
Using 3x10^8 m/s for speed of light, I make it 0.1 ms = 100 usec one-way to the balloon, i.e. 0.2 ms for ground-to-ground via the balloon. So the wireless part will be fine as far as Quake is concerned - it's just the other parts of the trip that may add latency.
In practice, latency may be much worse than this - look at GPRS, where latencies are many hundreds of milliseconds (often half a second) even though the wireless distances are much shorter. Wireless links use a lot of extra error correction and in the case of GSM/GPRS, packets are sometimes split into much smaller frames that are then interleaved with other frames - this is because GSM is voice-centric and this works well for voice. Not sure if GPRS has a better coding that avoids this, but if this balloon system is designed for data it should avoid some of the overhead.
GPRS will be everywhere that GSM exists today (i.e. available in almost every country), a fairly slow but always-available packet-based transport - it is sometimes unreliable at present at least in the UK, but it generally works OK and is fast enough for short emails and suitably small web pages. Great for killing time if you are waiting to meet someone and have exhausted your AvantGo offline reading.
GPRS will also be upgraded to EDGE in many countries, tripling capacity and improving performance a lot. W-CDMA/UMTS (i.e. true 3G for GSM operators) is much more expensive to roll out but will provide more flexibility and performance - the real issue is whether 3G will make enough money to pay back its investment.
802.11b/Wi-Fi is not really an alternative to GPRS - there is no way there will be hot spots covering the whole of a large city, which is what GPRS does today. Wi-Fi is great for use in hotspots and within businesses - the ideal is a phone or PDA that has Wi-Fi, GPRS and perhaps 3G, switching between them based on availability and cost. Nokia has already released a PC card that does Wi-Fi and GPRS, and one trend is to put GSM SIM cards (smartcards) into Wi-Fi PC Cards, enabling you to automatically bill your Wi-Fi usage to your GSM phone account.
Finally, this camera chip is of course not dependent on GPRS, it will work fine with any wireless technology including CDMA2000. Most of what I've said above applies to CDMA, with CDMA2000 1x instead of GPRS and CDMA2000 1xEV-DO instead of W-CDMA - the only difference really is that CDMA operators have a much easier upgrade path, and don't need much spectrum, but they will probably remain in the minority world wide compared to the 70% market share of GSM globally.
I think picture messaging (part of MMS [Multimedia Messaging Service] in GPRS networks) will be huge, and this camera chip will help it take off. If you can send a picture to anyone with an MMS mobile phone or an email account, you can send postcards to friends and 'how do I fix this' messages to suitable experts, and get 'top 5 goals' messages, photos from Internet personal ads, etc... The more people have a camera built into their phone, the more they will use it (though probably never as much as plain text SMS).
MMS phones are already available in Europe (Ericsson T68i, with Nokia 7650 soon) where MMS is just starting - in Japan, J-Phone has had a huge success with picture messaging, known as Sha-Mail (over 4 million picture messaging handsets sold). Watch this space...
Even if you don't have a mobile phone, you'll be able to send email with picture/sound/video attachments to anyone with an MMS phone.
Alastair Reynolds is very impressive - particularly Revelation Space, whose sheer scope is quite breathtaking.
You are exaggerating - I live in London too, and the trains and tubes are not that bad (at least from where I live). Calling the trains 'a death trap' is ironic when you consider that railway transport is one of the safest ways to travel - far more people die every year on the roads, but this is virtually invisible since it happens in many small accidents, compared to the large and well publicised rail accidents.
The real solution is time shifting and working from home - I commute into work at 9 to 9.30 am and the tubes are nearly empty.
This only works in Windows where you have identical apps everywhere. Unix workstation setups can have all apps installed on servers but run on the clients, meaning that an engineer gets a unique set of apps when they log on to the same workstation just used by a manager.
You probably have to see this to realise how much better it is than Windows.
GPRS also allows for billing by bandwidth usage - the real innovation in NTT DoCoMo's i-mode was the business model (NTT only taking 9% of total revenues vs 50% in Europe), which stimulated a lot of new content providers. Official providers bill based on content, via NTT. (I think CDMA2000 1x will also bill by bandwidth used, but it depends on the operator).
Other innovations that helped i-mode include the end-to-end design (handset, protocols, infrastructure, billing, business model, content selection and marketing all from a single company) - like the Macintosh, the result is an easy to use and attractive product, although quite proprietary. Another key factor is that many Japanese commuters spend 2 to 4 hours per day on overground trains, making i-mode ideal for killing time. There is probably also some truth in the idea that PCs and fixed line Internet didn't take off so fast in Japan, so i-mode phones were filling a gap, and also fitted into smaller Japanese apartments.
Some of these factors can be replicated in the US and Europe - I'm interested in trying something that actually works consistently and has usable and interesting content (i.e. nothing like WAP...).
Yes, the US phone system is so great that I regularly get just 20 to 25 Kbps on a 56K modem there, whereas in the UK it's always over 40 Kbps. (Irony alert...)
The best thing about the US phone system is the flat-rate for local phone calls, but that's also why cellular operators have to charge for incoming phone calls to mobile phones in the US (otherwise they'd lose money on every call). And this incoming call charging is a key reason why US residents don't give out their cell phone numbers, reducing the overall size of the cellular market and thus mobile penetration. This is the real reason for low mobile phone usage, it has very little to do with the quality of land lines, only their cost vs cell phones.
Japan is the only country where I saw ISDN data sockets on payphones everywhere, even tiny ski resorts, and voice quality was fine when I was there. If you really want to talk crap land line systems, try India - the GSM coverage is not too great but at least it works better than the very noisy land lines.
What would be really useful is a Bluetooth-connected Springboard dock, i.e. you plug your Springboard into a separate gizmo and Bluetooth networks it to your new organiser (Treo, Palm, Sony, etc). A bit slower probably, but OK for many applications, just needs software support in PalmOS to extend the Springboard interface over Bluetooth.
The only flaw in this idea is that Handspring have for some reason not put Bluetooth in their devices, unlike most new mid-to-high end phones in Europe, but Sony and Palm do have Bluetooth. They also forgot GPRS for some reasons - no idea why, since GPRS is available in every major European country and is rolling out in the US.
I have a Visor Deluxe but I just bought a Palm m515 so I can use Bluetooth to network to my GPRS phone (Ericsson T68). At the rate wireless technology is evolving, I think separate PDA and phone is the way to go - this also makes it easy to network a laptop to the phone, and laptop to PDA for syncing, etc. You can also use Bluetooth for wireless headsets for extra geek points...
You can already sign routing updates with BGP (using MD5 authentication) and probably with other protocols as well. BGP is the most important one of course because it is used between ISPs.
The Swedish warship, Vasa, also failed due to unrealistic timescales and lack of requirements validation. Many of these technology failures are really process/project management failures, of course.
I saw the Vasa in its museum the other week in Stockholm - they retrieved the ship from the bottom of the harbour and it is now on display, with very interesting exhibits about how it was built. Worth a visit if you are ever in Stockholm.
See http://panipcase.homeip.net - already posted by someone else but buried within a thread - this is a site for several companies already being sued by PanIP. They are all small companies of course, but if they club together and get help from the EFF etc, they could probably beat PanIP and countersue. It would be well worth signing up there and commenting on their discussion forum.
Mod the parent up - link to information on PanIP case...
Also, I'd recommmend contacting the EFF, as others have suggested - this makes it easy for people to contribute to the fighting fund.
Ultimately, the provider can always monitor how much bandwidth you are using by looking at its own routers - you can't spoof this. Search for 'Cisco NetFlow' for one example of how to do this.
By making it more expensive for them to detect cable modem uncapping, you are probably just going to encourage them to disconnect uncappers rather than just warning them.
GPRS, used by Voicestream, Cingular and AT&T Wireless, is a competing technology to CDMA2000 1x (i.e. what Verizon uses). I'm sure CDMA2000 services can be billed by the Kbyte, it's just that Verizon has chosen the older per-minute model. Every GPRS operator I know of (mainly in Europe, it's an upgrade of the global GSM mobile phone standard) uses per-Kbyte billing, to encourage people to stay on the network at all times. GPRS phones do have a 'connect' phase, but you can stay connected for days with no extra charges.
Anyway - the key point is that GPRS and CDMA2000 are both *technically* always-on, but may be implemented with a billing scheme that discourages always-on. Perhaps this is more common in the US, but IMO it's a big mistake. Once people are always-on, they can start doing ad-hoc messaging and web browsing whenever they feel like it, without the connection delay.
It is always-on, just not unlimited - like GPRS, which is always-on but billed by the Kbyte transferred.
This comes down to the service creation APIs that live on servers in the 3G network - while 3G has a huge range of such APIs, and a lot of capabilities in this area, it's rather doubtful that these will be fully exposed to the customer. UMTS Release 5 (a later release of the main 3G standards) will be wholly based on SIP, but you'll still need to go through a SIP proxy).
However, at least with EPOC APIs for C++ and Java you'll be able to fully program whatever features are exposed from the network.
I live in London, and the air quality here is appalling compared to most US cities I've visited - over there, I can actually *smell* the car fumes are cleaner...
Have to agree about transport though - despite the hassles of public transport in London, it is still easy enough to get around without a car.
Thanks for the pointer - Dreamhost look really good, and I've been researching web hosts for some time. Particularly impressive that they pre-install many optional Perl modules, include SSH and 100 MB space in the $10/month account, and have a very good knowledge base.
They look particularly good for more complex Perl CGI systems, e.g. TWiki, a web collaboration system I'm working on (see http://twiki.org).
It's certainly trivial to spend lots of money on a startup, but it's not essential. The programmers can work from home to start with, the execs can do their own admin, and so on. It doesn't help that the IDC report came out, but IMO running out of money that early is just bad planning.
Local calls are only 'free' in the US if you make enough to cover the monthly local calling fee (at whatever rate would be charged). I don't make a lot of local calls in the UK since I got ADSL, and just use my mobile for most social/business calls, so the US 'free local calls' scheme would actually cost me money.
The rest of the world is actually *bigger* than the US (shock!) and GSM works just fine there...
This is ridiculous - operators are NOT worried about this sort of thing. Have you thought about how hard it would be to modify a modern phone, with its incredibly tight packing of circuitry and large-scale integration, to be a scanner?
There are plenty of operators making money in Europe without needing to lock phones (though some of them still do).
Operators simply want to make it harder for people to switch, even though the subscriber is already tied in by a contract (meaning that the operator's phone subsidy is not a reason for phone locking - the contract ensures they will make enough money to cover the subsidy).
BT Cellnet (O2) and Vodafone have never locked phones in the UK, and nor do Virgin. Orange and one2one (now T-Mobile) do lock their phones. However, it doesn't cost much to unlock the phones, as long as your initial contract (typically 12 months) is finished.
Phone locking is fairly pointless IMO, and should be stopped - the fact that Vodafone, who are a very profitable wireless operator, don't do it shows that it's not necessary. After all, people are already locked in by the contract, so all you are doing is stiffing them for an extra 20 UKP or whatever to unlock.