Skip to the page

Plainly Put

One sentence about building things, taken apart

HomeExecution

Work on the future product as if it already exists

Treating a future product as real turns a forecast into a work plan. It also hides the difference between pieces that exist and pieces that do not.

See things in the present, even if they are in the future.

Larry Ellison

In 1970 an IBM researcher named Edgar Codd published a paper describing how data could be kept in tables and retrieved by what it contained rather than by where it sat on a disk. IBM treated the idea as research. A few years later three programmers in California read the paper and decided the relational database was a product that could be sold, although no customer had asked for one and IBM had not shipped its own. Their company became Oracle, and its first commercial release reached the market in 1979, well ahead of the company that had published the idea.

That episode is the best gloss on this line from Oracle’s co-founder. It turns up in collections without a date or a setting, so it is safest read as a compressed account of how Ellison worked rather than as a sentence from a known speech.

The instruction is practical. A founder who sees a future product “in the present” starts making decisions as if it were already real: hiring for it, pricing it, writing its manual, telling customers what it will do next year. The vision stops being a forecast and becomes a schedule. The alternative, waiting until the future is obvious, means arriving at the same moment as everyone else, usually with less money than the incumbent.

It also changes how obstacles look. If the product already exists in the founder’s head, the missing parts are engineering tasks with owners and deadlines. That framing carries a team through the long stretch when the thing does not work yet and nobody outside the building believes it will.

The same habit can mislead. In the mid-1990s Ellison argued loudly that cheap, diskless “network computers” would replace the personal computer, with programs and files living on servers. He described the shift as though it had already happened. It had not. For the rest of the decade PCs got cheaper faster than networks got reliable, and the network computer faded. The idea was not absurd, and much of it resembles the browser-based software that followed. The error was in the calendar, and in treating a conviction as a fact about the market.

That is the honest limit of the advice. Seeing the future as present works when the pieces that make it possible are already on the table and someone only has to assemble them, as with Codd’s paper and the minicomputers of the late 1970s. It fails when the founder is also waiting for other people to lay cables or change habits that have not changed yet. From the inside, the two situations feel identical.

A workable test is to write down what has to be true for the imagined product to function, then mark which of those things can be bought or built today. If most can, acting as though the future has arrived is simply efficient. If the list depends on customers or suppliers behaving differently first, the vision may still be right, but the date attached to it probably is not.