The Magazine

Apres Spam

The next email crisis.

Sep 29, 2003, Vol. 9, No. 03 • By DAVID GELERNTER
Widget tooltip
Single Page Print Larger Text Smaller Text Alerts

You see an email, plan to respond but forget. Especially likely when the email is long or complicated. By sending a quick acknowledgment, you give the sender permission to nudge you (in a reasonable way, after a fair interval). The conversation is less likely to flicker out by accident.

You finally remember to respond, but you've forgotten the details. It suddenly hits you that you intended to answer a message from somebody about some piece you once published--but a message from whom about what? A conscientious correspondent will shoulder his virtual shovel and dig the thing out. In practice, the exchange is probably going nowhere unless the sender decides to try again.

You ignore an email on purpose. In this case you are better off without the protocol. Under the protocol, you will be forced to ignore not only the original but the duplicate. Reform is rarely cost-free.

These rules could be put into effect right now, using any mail system. But they are pointless unless a whole community uses them, and would be a nuisance to apply using conventional mail software.

What's required is a two-button mail-reader. One button is labeled "acknowledge (quick!)," the other "answer (slow!)."

When you press the "acknowledge" button, you see a time-ordered list of all messages you have yet to acknowledge. Also included: repeat messages that await resending--you've sent them once but they have not been acknowledged, or remain unanswered. And acknowledgments received--just to glance at; no action needed.

Maintaining such a list is fairly complicated, but software does the work. Whenever a new message (except an acknowledgment) arrives, the system will ordinarily start a 24-hour timer, put the message on your "acknowledge" list and prepare an acknowledgment to be launched when you are ready. Whenever you kick off a new conversation, the system will start a seven-day timer; when the timer runs out, your original message is hauled out of storage, readied for relaunch, and added to the "acknowledge" list. All these manipulations go on behind the scenes. Pressing "answer" shows you messages you have acknowledged but have not yet responded to.

Everything on "acknowledge" can be dealt with quickly. If you check it several times a day, you can keep up without bogging down. You might deal with "reply" once a week. Bottom line: You keep afloat by relying on mass-processing efficiencies (your time-consuming correspondence is bundled into a neat batch to be dealt with every now and then), and by discreetly relaunching stalled conversations without being obnoxious. Which is exactly what nearly everyone tries to do anyway. But your two lists let you do the obvious thing at minimal cost in overhead, and without driving your correspondents crazy.

I have left out all sorts of difficulties and special cases--and not only because they are boring. They are also pernicious. The software industry tends to start with the details, difficulties, and apparent constraints, and to work timidly upward from there. Complex mediocrity is the usual result. To design software that is simple, powerful, and general (or anything that is simple, powerful, and general), you must start with the overriding goals and the big picture and work downward, forcing technology into the mold you have decreed instead of letting it sprawl slime-mold-like into any shape it naturally favors.

ULTIMATELY we can't go on like this. More and more of the world's business is going online. The online digital universe, the "cybersphere," is turning into a mirror world where every real-world entity has a software doppelgänger. Today's conventional software has no way to cope with such a development.

But there is a way to counteract ever-higher volumes and varieties of online information: by making the interface far simpler and more uniform. Every digital item you own or ever will own will be stored in a single structure. (Various companies, including one I work for, are building this type of software.) This single structure with all your information inside will be accessible from any computer or quasi-computer anywhere. (Any cell phone, laptop, answering machine, TV, automobile.) It will be easy to display, to visualize, to manipulate. Thus, a sort of "information beam" that grows brighter all the time (as more and more information is added), but can be focused easily with pinpoint precision. To handle rich, varied, and voluminous information, you need a simple and uniform package. The book (the physical object--sheets bound on end) is the finest design in history for exactly that reason. A book might be about anything, but all books work the same way. When software design is a tenth as sophisticated as book design, we will be getting somewhere.

David Gelernter is a contributing editor to The Weekly Standard.