In the late 1970s, I was thrilled by the ability to send and receive messages through the revolutionary medium then known as “electronic mail.” At about the same time, I began to write my letters, notes, articles, and books—we didn’t yet call all such things “documents”—on a computer and store them electronically rather than on cards or papers stuffed into filing cabinets.
Since those days I have thought of the information aspects of life as an unwinnable race. With each new year and each new Moore’s Law–enabled boost in processor speed, transmission rates, storage capacity, and other elements of the information infrastructure, ever more data has come at us, ever faster. Appointments, assignments, driving directions—everything has become part of the expanding cloud of “personal information” necessary for our work and home lives.
Year by year tools for coping with the onslaught have also been improving—but more slowly. Electronic calendars, online collaboration systems, search engines and archives, and the converging technologies that make up the smartphone revolution did something, but never quite enough, to put people in control of information. “I sometimes wonder whether, with all this data, people have just given up,” Mitch Kapor, a personal-software pioneer, told me recently. “They may just have resigned themselves to living in this infinite sea of information.” Kapor first became famous as the founder of the software company Lotus and the designer of the spreadsheet application Lotus 1-2-3, but I revere him as a creator of the brilliant early personal-information program Lotus Agenda. (Agenda, which I wrote about in The Atlantic back in 1992, is a relic from the MS‑DOS age. All these years later, I keep DOS emulators on my sleek, modern MacBook Airs purely so I can run the program from time to time.)
Tech entrepreneurs don’t usually sound downcast, and in fact Kapor’s overall vision of where we’re headed is upbeat. He is one of the five experts I asked to speculate about the future of personal-information technology, especially whether the race for mastery of one’s own data might someday seem winnable. The others were Esther Dyson, a longtime technology investor and analyst; David Allen, the originator of the Getting Things Done productivity system; Phil Libin, the founder and CEO of the software company Evernote; and Mark Bernstein, the chief scientist of Eastgate Systems and the designer of Tinderbox, an information-management program. Individually and collectively, their comments boiled down to: We’ve been through the worst. The next stage in information technology will put people back in control, or closer to it. More specifically, our conversations foretold:
1. The beginning of an end to the e‑mail nightmare
E-mail is indispensable, and unendurable. That is because it does not scale. Every message, as Esther Dyson has written, “represents a task—something to read, a query to answer, a meeting to schedule, a bill to pay, a request to fulfill or deny.” Thus senders can generate more tasks than recipients could possibly perform. As she told me, “The reader’s time is free to the sender, which is a huge market inefficiency.”
Dyson says that some market mechanism will reset the balance. One way or another, senders will pay a premium for recipients’ time and attention—as they did in the pre-e‑mail days, by having to request appointments or make sales calls or, at the very least, pay for postage. Phil Libin says improved filtering systems are already solving the problem. “I have 100,000 e‑mails I haven’t answered,” he said. “I know that I can’t even open 90 percent of the e‑mail I get. Am I missing something important I should see? Sure, but rarely.” The remaining challenge is to reduce “the error rate”—that is, the share of important e‑mails that he does miss. And this, Libin said, should be “an easily solvable” problem, with the help of systems that learn whom he wants to hear from, and whom he doesn’t.
2. The spread of anticipatory intelligence
Computers work best when you’re least aware that they’re working at all. Modern cars, for instance, contain the processing power of dozens of early mainframe computers. But most of the time they discreetly scan for problems and alert the driver only if they detect something—engine trouble, poor traction on slippery roads, low gas—the driver might need to address. Someday, car computers might even be powerful enough to detect texting at the wheel on the basis of irregularities in driving patterns.