Throwing Stuff Out is Obsolete
From time to time, I tidy up my hard disk. I delete useless old files and excess Tinderbox notes. I weed out the worst snapshots, the redundant images and the blurred pictures and the pictures I accidentally took of someone's feet.
This is probably a mistake. It often costs more to decide to throw something away than to save it forever.
- I can buy a terabyte firewire drive for about $1000. That's not the optimal price-performance, but the numbers are nice and round. That drive ought to last at least 3 years. That works out to $1/Gb.
- Even a very large grocery list is going to be less than 100Kb. But we'll allow 1Mb, just to be liberal.
- The "rent" of storing an extra 1M on our hard disk, for three years, works out to about a tenth of a penny.
- If Moore's law holds out, in 3 years our $1000 will buy us a 4Tb drive. And 3 years from then, it'll be a 16Tb drive. Say 'hello' to our old friend, the geometric series. 1+1/4+1/16+.... This never gets above 1.333. So, if Moore's law held forever, we'd never spend as much as a penny to save the grocery list.
- Even if progress stopped immediately and storage costs never declined, the rent for 300 years of drive space is only a dime.
In practice, storage costs will be dominated by the time you spend maintaining the system, copying files every three years to a new disk, and pruning unwanted files.
This has interesting implications for software design. Frank Shipman's VKB, for example, saves your undo stack forever, so you can go back and look at the state of your document last October. You never know what you might need.