Fog Creek Software
Discussion Board

Platform Business model

Ok, some thoughts and question here about "what's the business model for developing a platform?"

For MS Windows, it's obvious now.  They get money for their OS from the end user (hidden in the cost of the computer).  How long did it take for this model to come to fruition?  Anyone know how long it was between a freely redistributable Windows 1.0 and when people started to pay for it?

Sun is supposed to get their money for Java from - where was it? - hardware JVM support?  Anyone know how this is working out for them?  But at least this one I understand, since they don't have to give away what they hope to sell in the future.

So Groove should give away their sole product until it becomes popular enough for people to want pay for it?  How long will that take?  And does anyone think that by the time it matures to that point, there won't be a free Gnoove (I get credit for the name if it takes off!) platform ready to compete with it?  Why or why not?

How should a company move from producing (creating, selling and marketing) an application to producing a platform?

Friday, August 30, 2002

Application vs. platform doesn't have to be a binary choice. Groove is really two things:
- an implementation of p2p architecture
- an application that uses this architecture to provide useful things (Groove desktop thingy that contains various groovelets like file sharing, collaborative notepad etc.)

Groove could and should allow people to easily use their p2p architecture. That could be separated out of Groove as a set of redistributable dll's that app developers could bundle for free with their apps.

Groove would still sell their own app the same way it does now.

The problem for Groove Inc. with that model is that it lowers a barrier to entry for competing with Groove app. So they might have a smaller share of the "apps using Groove p2p architecture" market. The opportunity here is that it creates a bigger market for everybody so Groove might end up netting more. Competition also drives innovation. Groove has a head-start but they would probably have to try harder (I played briefly with Groove and was very disappointed with the lack of features of groovelets; those were almost toy apps). Usually competition works well for everyone involved.

If they don't push their platform, someone might get fed up and re-create their p2p architecture on better terms and that might get more momentum than Groove p2p and then Groove will be in serious trouble.

I have additional beef with Groove ( ): it has state-of-the art p2p architecture but the apps they've built so far on top of that architecture are third-rate.

Krzysztof Kowalczyk
Saturday, August 31, 2002

This is a very very excellent topic.

It is often assumed that once your platform is popular money magically follows, but is that the case? We should really examine some platforms here.

Windows - makes money by selling the software itself, but mostly by making applications on top of it. MS has and advantage over other because it steers the platform, and has the earliest access and experience with it. It uses this to create new standards like office that aren't open. The platform is mostly bundled with new machines at a cost, but somewhat invisibly. Hardware vendors have been more or less forced to include it in the past.

Swf - Open format, valid platform. They make money by selling the only legit tool to make swfs. Others who try aren't given the full spec, and more importantly, aren't given the upcoming spec until the the new version of the tool is released. So things like scripting enhancements will always be one version behind. It is given away via the net, and lots of work is done to make it installed on every browser/device/os etc. There have been failed attempts to get revenue out of server tools for it in the past - most revenue still comes from the tool. There are also sales due to bundling related products like dreamweaver/fireworks, which all share a common interface.

Pdf - I'm not certain of the openess of this, or revenue - there is the tool, and then potential licencing from ebooks or publishers? Helps to sell related adobe products too, common interface too. (in fact a lawsuit to stop M from using its interface, so maybe that is key here)

MacOS - Lock on hardware and OS, competes in software with an advantage like windows does. Arguably a more closed platform, but nothing like things like playstation etc.

Playstation - probably loses money on hardware, but gets a cut on software. It also totally takes developers to the cleaners, charging 100,000+ for the equipment and software needed to make a game (it kind of refutes the 'give away the sdk theory, no?'). It also has very very strict control on ok'ing a game, standards, quality etc. Still at the top of the game market, and started from zero in a market dominated by others, so you can't ignore this model.

Java - give away the platform, and the dev tools, and quite a bit of control. Not sure about revenue...?

.Net - give lots away, sell the tool, own the future OS, but give up some to standards like ecma...? Obviously this is also trying to be a circle play around java, so it isn't just revenue, though there are revenue streams in their obviously, as it is clearly an MS centric thing.

Rambus - set an open standard and then secretly patents parts of it ; )

others? corrections? failed platforms?

I think if you are planning to give away your platform, you better have a clear idea of how that will bring revenue. I don't think its a given that a popular platform will make you cash.

Robin Debreuil
Saturday, August 31, 2002

"I don't think its a given that a popular platform will make you cash."

It certainly isn't. But it does give you a tremendous edge and in extreme cases it can *help* you to achieve astronomical, unheard of success just like Windows *helped* Microsoft. I emphasize *help* because, as you said, it really isn't a given. If Microsoft was otherwise lousy company it would end up like Commodore Amiga (which commited a crime of not investing in the platform and making it difficult for people to develop). There's no free lunch but having coupons certainly helps.

But look at Netscape, they had extremely popular platform and failed spectacularly.

Krzysztof Kowalczyk
Saturday, August 31, 2002

While maybe the windows runtime became free, the platform it was running on was not (dos). Hence, the giving away of windows runtime was certainly to help people use windows, and at the same time it made the dos platform more valuable.  Today, they give away DOS, and you now pay for windows!

There is also something else very important here. While you can give, or make some tools away for free, there needs to be one other ingredient here.

There is a great lesson that MS’ realizes here, and even Eric Raymond in the Cathedral and the Bazaar points out.

In Eric’s book (Cathedral and the Bazaar) he points out that simply deciding to create and release a open source project will not take on a life of its own unless it has some functional use already. In other words, a open source project will not succeed unless the *EXISTING* product has some use, and users will adopt it.

This same lesson actually applies in the business world of software. Joe’s example of the JET engine is a perfect example. JET has several things going for it.

1) It is a very capable little database engine
2) It is free
3) (and this one is most important) – EASY to use.

Microsoft for the last few years has been trying get developers to drop the JET engine, and adopt the free sql-server engine. This newer database engine is also free, and it is 100% compatible with sql server. Not only is the server engine free, but it has all kinds of additional features over and above JET, such as server side code, triggers, transaction playback etc.

Why then is no one using this fabulous and free database engine? (we are talking about the free MDSE engine from MS?) In fact, I see more people using MySql with Microsofts ’own products over MDSE! Why?

Out of the above 3 reasons, #3 is the problem. I have in the past installed MySql, and also the MDSE on my pc.  It was a much better experience when I installed MySql (so much for that myth about open source stuff not being easy). Not only that, the MySql had all kinds of stuff included to get me up to speed. Right after you load MySql, you can start playing with the engine. It is fun, and rewarding right out of the box.

When you load the MDSE, it just sits there, and you can’t play with it. Why oh why did they not package the Enterprise tools with the MDSE I will never know. The MDSE could be a wild success with us developers, but we just don’t bother, since no other tools are bundled with it...big mistake. (MS has the right idea...but bad execution here!).

So, in addition to giving away the “runtime” or the free engine....don’t forget to give some functionally also.

This applies to both platforms or software. Hence, that is why Joel does gives away City Desk, and it has full functionality. A great way to get us all hooked.

Platform or got to hook em....

Albert D. Kallal
Edmonton, Alberta Canada

Albert D. Kallal
Saturday, August 31, 2002

I don't think Sun ever had a "strategy" for Java beyond "this is something really neat, and it'll mightily piss off Microsoft". That said; as with any good idea, Java has found its niche and Sun is in a great position to exploit it, being the creator and holder of all the keys.

Cross-platform, cross-appserver-portability hype aside, the real Java money is coming from bundling a triumvirate in one package. Big expensive hardware, big expensive application-server software, and big expensive consulting bills to turn the first two into big-business applications. This is where IBM is right now with its Java offerings. Selling its hardware range, plus the Websphere suite of products, plus Global Services (which just consumed PWC). This is where Sun is going with its hardware, and SunONE, and strategic partnerships for the consulting side.

And they're doing it quietly and quickly. Sun were _nobody_ in the J2EE market six months ago. Last month they replaced Websphere as the Java technology of choice at Australia's biggest telco, and they're pushing aggressively into the banks.

Expect, in the next year or two, the final death of BEA as a force in the Java market. BEA are irrelevant now, they just haven't quite worked it out yet. In the really-expensive enterprise sphere, Sun and IBM will duke it out with each other, and with .NET. Down below, Orion and its pals will continue to operate in the low-budget range, competing with open-source projects like Tomcat and Jboss.

Saturday, August 31, 2002

So, in the desktop software market, is Windows the only platform for which the user pays?  It sounds like most of the others mentioned (flash, pdf, java, .Net) the platform is given away for free, and revenue comes primarily from developers.
What I find interesting is that the whole anti-trust case against Microsoft is based on the fact that their platform has a dominant position, and that they use the platform to leverage their applications that run on the platform (office, IE, etc).  But isn't this essentially the strategy that Joel advocated?

Tuesday, September 3, 2002

*  Recent Topics

*  Fog Creek Home