Let Java Go

An open letter to Scott McNealy, CEO of Sun, responding to remarks at Sun's February 2004 analyst meeting.

A Dutch translation is available,

The open-source community has been hearing reports that you have recently said of Sun Microsystem's strategy "The open-source model is our friend". We're glad to hear that, and Sun's support of OpenOffice.org certainly puts some weight behind the claim. But that support is curiously inconsistent, spotty in ways which suggests that Sun is confused in the way it thinks about and executes its open-source strategy.

That confusion is evident in another of your quotes. Many of us think you are right on when you say that "Sun [...] is less threatened by a zero-revenue model for software than just about anybody out there." We agree that the potential for you in using open-source software as a value multiplier for Sun's hardware business is huge. This wouldn't even be a novel move for Sun; your release of the NFS standards in 1984 was possibly the single most successful market-shaping maneuver in your company's history, and we'd love to help you repeat it.

But the casual equation between "open source" and "zero revenue" suggests that on another level you don't really know what you're talking about. Open source is hardly a zero-revenue model; ask Red Hat, which had a share price over triple Sun's when I just checked. Or ask IBM, which is using Linux as a lever to build a huge systems-integration business in markets like financial services that Sun has historically owned.

It doesn't have to be this way. If Sun were prepared to go all the way with open source it could seize back its position of industry leadership. Sun is one of a small handful of companies that would both have the smarts and the street cred to do even better than IBM has from a full-fledged alliance with the open-source community. Indeed, on historical grounds you might do better; many of the senior people in the movement are old-time Unix hackers who remember that Sun was founded by geeks like us at a time when IBM was the Great Satan.

But Sun has done other things that make us wonder if the vision and courage to choose the open-source path are really there. The suspicion persists that OpenOffice.org is just an expedient way to poke Microsoft in the eye, not the cutting edge of a open-source-friendly strategy that will position Sun for the future. Matters aren't helped by the fact that Sun appears, with Microsoft, to be one of the two companies doing most to stuff SCO's war chest for its attack on Linux.

In 1987, three years after the success of NFS, Sun lost the war to define the standard graphics interface for the next generation. The winner, the X Window System, was technically inferior to Sun's NeWS offering. But X had one critical advantage; it was open source. Ten years later in 1997, when Bill Joy came to a Linux conference to push Jini as a universal network-service protocol, we in the open-source community told him straight up "You can have ubiquity or you can have control. Pick one." He picked control, and Jini failed in its promise. The contrast with NFS could hardly be more stark.

Today, the big issue is Java. Sun's insistence on continuing tight control of the Java code has damaged Sun's long-term interests by throttling acceptance of the language in the open-source community, ceding the field (and probably the future) to scripting-language competitors like Python and Perl. Once again the choice is between control and ubiquity, and despite your claim that "open source is our friend" Sun appears to be choosing control. Sun's terms are so restrictive that Linux distributions cannot even include Java binaries for use as a browser plugin, let alone as a standalone development tool.

Mr. CEO, tear down that wall. You have millions of potential allies out here in the open-source community who would love to become Java developers and users if it didn't mean ceding control of their future to Sun. If you're serious about being a friend of open source, if you're serious about preparing Sun for the future we can all see coming in which code secrecy and proprietary lock-in will no longer be viable strategies, prove it. Let Java go.

Eric S. Raymond
President, Open Source Initiative
12 Feb 2004

I have written a followup to this letter.

Note: This letter has attracted a lot of half-bright criticism to the effect that I should have compared Sun and Red Hat by market capitalization rather than share price. Wrong. Market cap measures present value; the relative prices investors are willing to pay for shares (not the company's asking price) reflect their expectations of future return per their invested dollar (that last qualification is what normalizes share volume out of the equation). I'm informed that this effect is particularly visible in IPOs and derivative trades, though my informant thinks prices of established stocks are too noisy for it to be reliable there.