I went to a garden party last weekend. It was the kind of party that is everything Silicon Valley promises to be to anyone in England wondering about living the "California Dreaming" lifestyle. Powder-blue endless skies, gentle golden sunshine, a live band playing non-intrusive jazz standards, exquisite canapes, and good conversation with a range of people from all walks of the technical and business side of the Valley.

Most were wealthy of course, but then again in this place anyone who owns a house is probably a millionaire. But even by those standards these were the very wealthy, with tens and sometimes hundreds of millions of dollars to play with. The venture capitalists (VC's).

David Icke was right - the people who rule the world are seven foot tall lizards. But like the Slitheen family on the new Doctor Who they hide inside Silicon Valley venture capitalist human suits, living up there above Stanford University on Sand Hill Road in their own strange little world. Occasionally they come down amongst us mortal engineers and even grubbier Free Software programmers to dispense wisdom and mountains of cash to fortunate start-up companies. I was the only Free Software advocate there of course. They tolerate me, like the trained monkey they're expecting to have their pictures taken with after dinner.

Pills and spills
"A startup company is like a bicycle", said one VC to me. "The two wheels are marketing and engineering and you don't get very far without both", he finished. "Yeah, but not just any bicycle" I replied bitterly, "It's a penny-farthing. The front wheel is marketing and is the large essential part. The back wheel is engineering. Small, insignificant, and when it breaks it's easy to replace it with another one". Just then the head of a company selling "E-mail direct marketing services" (aka SPAM) interrupted. I asked him how many viagra tablets he'd sold this month and the subject of the conversation changed to less serious matters.

But as jaundiced as my view sounds, I actually have experience to back it up. Back in the pre-Dot Com boom year of 1997 I had the great fortune to have the choice between two equally matched start-up companies. I haven't changed the names in the vain hope of helping to punish the guilty.

Hot
Start-up number one was Whistle Communications. Whistle had one of the best engineering groups I'd ever met. Staffed by ex-FreeBSD hackers and GUI engineers from Apple, they had created one of the most exciting products I'd ever seen. It was called the Whistle InterJet, and was a one-stop embedded box to allow small businesses to get themselves a presence on the new, exciting medium called the Internet. It had the easiest set-up that was possible for an Internet device. It was designed to be purchased from your Internet Service Provider (ISP), and you set it up by connecting it to the phone line and typing in a serial number that the ISP would helpfully provide to you when you bought the Internet service from them. It configured itself automatically from there onwards. The end-user administration was completely browser based, a revolutionary idea at that time. It had a built in ethernet hub and even an option to purchase it with an ISDN adaptor (no one knew DSL would be popular back then). Samba was used to provide the means for the small business to create Web pages on their Windows client PC's and then copy them to the InterJet where they would be instantly mirrored onto the ISP's more powerful servers.

Cool
The other start-up was Cobalt Systems. They had a square box molded in blue plastic. It ran a derivative of Red Hat Linux on an underpowered MIPS processor. But it looked cool. Really cool. As an example of how cool it looked, one of the hottest restaurant spots in Silicon Valley, the Tied House in Mountain View, placed a Cobalt blue box in their office windows, which most of the patrons walked past on their way into the restaurant. It looked like a lifestyle accessory in the same way that an Apple iPod does today.

The software on it was nothing special. Just Linux, with a web server and the basic tools, Samba and the like (which was why they wanted me). The problem I had with it was I just couldn't see what Cobalt had that any other company couldn't create just as easily, faster and cheaper. They had no engineering differentiation I could see. I chose to join Whistle. I must admit I had great fun working there. Every night at 8pm the engineers would down tools and we'd have a rip-roaring GPL vs BSD licensing argument. Ah, happy times.

But you all know how this story ends.

Blow your own whistle
Cobalt creamed Whistle when it came to actually selling boxes. The InterJet won all sorts of awards. Technical magazines pronounced that it had even created it's own new category of product. Still no one bought it. Frustrated potential customers trying to buy the box from uninterested ISP's dubbed it "the nicest box you couldn't buy". That's when the Whistle marketing people realized too late that ISP's had enough difficulty selling their own product let alone co-marketing someone else's.

Cobalt on the other hand experimented with risqué advertising, rack-mounting (still with the same cool blue front panel) and other creative marketing techniques. Whistle compounded the errors by using a Windows NT based hosting service to serve the company website to customers. We frustrated engineers almost resorted to violence to get our marketing people to change to a FreeBSD hosted solution so customers would at least believe we had confidence in our own web-serving technology. Marketing refused: "We're not our own target market", they kept bleating. Our competitors told our customers we didn't trust what we sold enough to use it ourselves. We became a laughing stock.

The wheels fell off the bicycle. As I left Whistle, they were preparing to sell themselves to IBM, who were interested in the product as it had done tolerably well in Japan, where the monopoly nature of the telecoms industry and the fact the InterJet was supported by NTT (Nippon Telephone and Telegraph) removed the problem of apathetic ISPs. Whistle was swallowed whole by IBM, who later canned the product probably as part of some "seventh-tier reorganization of the department of Information Adjustments".

On your bikes
Cobalt didn't quite dominate the market they way they'd hoped, but did so well that they eventually suffered just as sad a fate (for their wonderfully talented marketing people at least) and were bought out by Sun Microsystems. But at least the employees were able to cash out with big stock option gains before Sun folded the company.

Even having Kirk McKusick, of original BSD UNIX fame, doing engineering work on the InterJet operating system didn't save Cobalt fromn the knacker's yard. This should be a lesson to prospective Free Software and Open Source companies, and also their evangelists. Sometimes it's not the polycarbon alloy the wheels on the bicycle are made of that matter, it's the exciting picture on the box it comes in.

We in the Free Software community are great at doing engineering, now we need to get serious about the marketing. As someone I intensely disliked when I lived in England once said, "Get on your bikes". Go fix it.

Jeremy Allison

Leave a comment below or send feedback to editor [at] tuxdeluxe.org