Consilient Network· The graph of everyone who built this · Add your entry ↗
The thread

Software as a Service

One idea, arrived at twice in the same family, twenty years apart, and built by other people both times. It is recorded here because priority claims are worth stating plainly and because the architecture this site describes rests on the same insight carried one layer further.

1975 · a basement in Washington1995 · Evolve Software2006 · the AWS announcement
Part I

1975 — the basement

Bill Whitson retired from the government, the military and everything adjacent to them, and started a company in his basement. It ran a hosted application for managing liquor-store inventories.

He had concluded that it was foolish for companies to own computers and write their own applications. So he bought a number of machines — ruinously expensive at the time, consuming the entirety of his daughter's inheritance — and ran the software for other businesses. There happened to be a liquor store near the house.

What he had actually built

This was the mid-1970s. Persistent storage meant tape drives, a substantial advance over punched cards. And the model he had arrived at was the application service bureau: a platform on which many businesses run software they neither own nor operate.

The application was not the point. The platform was. That distinction is the whole of the idea, and it took the industry another quarter-century to act on it.

He died of a stroke while playing tennis, in his sixties, before anything came of it. Under today's cardiovascular care it would have been a minor scare.

The rest of the Whitson record →

Part II

The nineties — rediscovery

Hare arrived at the same model independently, after his experience at Tesseract, and argued for it at Evolve Software under the name it then had: Application Service Providers.

The analogy he used was the payroll service bureau — ADP, Ceridian — businesses that had long since worked out that companies would rather rent a capability than build and operate one. There was no reason the logic should stop at payroll.

What it was supposed to fix

The consequence he argued for was not primarily technical. It was commercial: the elimination of long sales cycles and slow implementations — the enterprise software model in which a purchase took a year to negotiate and another year to deploy, and in which most of the value was consumed by the deployment.

He worked with Clare Gillan of IDC, who published on the topic and helped give the category a name.

Part III

The conversations

Hare and Marc Benioff discussed the model repeatedly in the late nineties, at a regular Friday-evening gathering at Enrico's in North Beach, arranged by their mutual friend Russ DeLeon — Hare's in-house counsel at Evolve.

They had first met when Benioff released Oracle Card, which Hare considered a trimmed-down clone of Apple's HyperCard: derivative, quickly built, and an early instance of what would later be called agile development. Hare was at Blyth Software at the time, whose product Omnis he thought considerably more robust, and he was frankly puzzled about what Larry Ellison saw in the man.

What was discussed, and what followed

What they talked about, repeatedly, was Hare's longstanding argument for Application Service Providers and what it would do to enterprise software economics.

Benioff went on to build exactly that at Salesforce, and to brand it with one of the most effective marketing campaigns in the history of the industry: the End of Software. Between roughly 2003 and 2012 Hare met dozens of people at Salesforce across several generations of executives — including Benioff's co-founders from the Telegraph Hill apartment — under Benioff's sponsorship and continuing friendship.

On credit

Benioff and Bezos have earned their yachts and everything else. They deployed at global scale, which is the harder job and the one the world actually needed doing. The claim made here is about where the idea came from, not about who deserves the proceeds.

Part IV

2006 — the AWS announcement

Hare spoke on the morning technical panel at the conference where Amazon Web Services was first announced — the event that had begun as a peer-to-peer and asynchronous agents conference at the St. Francis in San Francisco, co-hosted by Tim O'Reilly and John Battelle, and was later renamed Web 2.0.

The other panelists included an Amazon evangelist, a friend who had co-founded Macromedia, and a Yahoo executive whom everyone assumed was the star of the panel — on the reasoning that Amazon sold books, and what could a bookseller possibly know about software infrastructure.

What he said at the time

Hare called AWS the best thing he had seen in a decade. It was the future, and it was also the past.

It fulfilled the vision he had been arguing for since Evolve — and it was the model his grandfather had built in a basement thirty years earlier. Without it, Amazon is a very large online bookshop.

Pending confirmation

The details of this panel — the venue, the co-hosts, the other participants — are recorded from Hare's recollection and have not been independently verified. Anyone who was in the room and remembers it differently is invited to correct the record.

Part V

What is and is not being claimed

Stating this precisely matters, because a vague priority claim is worth nothing and an overstated one discredits everything around it.

Claimed

Independent arrival

Bill Whitson built a hosted application service in 1975. Hare arrived at the same model independently in the nineties and argued for it publicly and repeatedly before it was built.

Claimed

Documented advocacy

The ASP argument at Evolve, the work with Clare Gillan at IDC, and the conversations with Marc Benioff predate Salesforce.

Not claimed

Authorship of what was built

Salesforce and AWS were built by other people, at a scale and a level of execution neither Whitson nor Hare attempted. That is the harder half of the work.

Why it is on this site at all

Because it is the same insight, one layer down. Software as a service removed the requirement that a company own the machine in order to use the capability. This architecture removes the requirement that anyone surrender a resource in order for it to be used at all — and settles value back to whoever made it possible, which is the part neither the basement nor the platform era ever had a mechanism for.

It is also the clearest instance in the record of the problem the architecture exists to solve. Bill Whitson's idea produced enormous downstream value and no attribution, because there was no mechanism capable of recording what a contribution caused. Neither did Hare's. That is not a grievance; it is the specification.

Correct or extend this record → The architecture →

← PreviousOrigins
+ Add to the record