How Technology Keeps Failing Us
and why we keep
repeating our mistakes
A look at the sidelines of history, to many of the research projects and products that promised to change how we use technology, but ultimately failed to make a big change in our everyday lives. Nostalgy for the lost opportunities, reminiscing about the promised futures, and an insistent hope that one day some of these things will come true, in one form or another. Hopefully these old new ideas will spur new innovation in you.
The topics covered will mostly be in the realm of computers and electronic gadgets, as that is what I've fiddled with.
Talk at BIL 2009-02-08, 15 min.
Sidelines of history
that never made it big
but should have
(I think so!)
15 minutes isn't that long, so I'm going to concentrate on one particular area where we see some struggling attempts to solve the problems, but where I would like to see a lot more emergence of functionality, something greater than the individual features.
15 min time
- talking about just one scenario
- grab me in the hallways to talk more
Plan 9 from Bell Labs
- Unix rethought
- everything is a file(system)
- seamlessly distributed
- "infinite" storage
- Write Once, Read Many
- optical storage, before CD-Rs
- hard drive was cache, written to WORM every night
- same idea, just hard drives
- old versions of
- don't bother with backups
- eternal: growth was slow enough
- move apps to server; no desktop IT
- maybe force standard desktop
- "stateless desktop"
- your keycard on any workstation
- anytime, anywhere
- easy, convenient, fast
- everything exactly like you left it
On-Demand OS Image
- NFS root
- Live CDs (also VM, rootz)
- OS Circular: full network boot
- read-only root filesystem over the network
- "no install software", a read-only image fetched over HTTP and loop-mounted, with writable tmpfs union
OS Circular: http://openlab.jp/oscircular/
- project to run virtual machines backed by SHA-verified blocks fetched over HTTP
- move apps & data onto laptop for travel
- Coda allowed offline use of network filesystem
- block-level migration for non-shared storage
- virtual machines can do live migration
- Android makes apps seem "always on"
- migrate virtual machine from server to laptop, leave snapshot on server in case laptop gets lost
Also interesting: http://en.wikipedia.org/wiki/Crash-only_software
What all that gets us
- apps can migrate to faster CPUs
- backups for everything
- you can use any computer you choose to trust
What went wrong?
- throw-away research projects
- people believing in Imaginary Property; greed
- short organizational memory
- too much productization, too little standardization
Slides are at eagain.net