Submitted by Justin Klein Keane (not verified) on December 8, 2006 - 4:27pm.
A couple of other thoughts have occurred to me over the day:
1. Privacy. As soon as you rely on network enabled applications you are entrusting your data to someone else (even if only for a short time). Also, data is exposed on the wire unless you encrypt it which increases overhead (and even then it's not necessarily safe).
2. Hardware connectivity - browser models carefully insulate the hardware from the network because the network is an untrusted medium. As soon as you start talking about doing things like synchronizing your PDA with your data you move beyond the web browser because it cannot be trusted in this hardware model.
3. Offloading computation to a server is a tempting solution to many problems, but it runs into issues. While you might say "let Google's servers do your processing" I would reply "use open source and let your $300 machines do the processing because a) you already own them and b) it's totally free". Using open source software I can deploy an entire business for less than one that uses managed/hosted solutions. Moving the processing to the server means that you have twice the cost - once for your hardware, and once for the hosts hardware. You're lining that cost up in a column next to the price of software - but when the software is free that calculation skews in the other direction.
Yeah but...
A couple of other thoughts have occurred to me over the day:
1. Privacy. As soon as you rely on network enabled applications you are entrusting your data to someone else (even if only for a short time). Also, data is exposed on the wire unless you encrypt it which increases overhead (and even then it's not necessarily safe).
2. Hardware connectivity - browser models carefully insulate the hardware from the network because the network is an untrusted medium. As soon as you start talking about doing things like synchronizing your PDA with your data you move beyond the web browser because it cannot be trusted in this hardware model.
3. Offloading computation to a server is a tempting solution to many problems, but it runs into issues. While you might say "let Google's servers do your processing" I would reply "use open source and let your $300 machines do the processing because a) you already own them and b) it's totally free". Using open source software I can deploy an entire business for less than one that uses managed/hosted solutions. Moving the processing to the server means that you have twice the cost - once for your hardware, and once for the hosts hardware. You're lining that cost up in a column next to the price of software - but when the software is free that calculation skews in the other direction.