Scale and Competition on the Internet

Tom Foremski has a must read article on the Google YouTube acquisition. He's one of the few who get the big picture.

Lets say you want to hit a home run in a start-up. If you shoot too small, you end up with a Facebook - a "nice small business" (Jordan Rohan, an analyst for RBC Capital Markets). Too big, and then everybody else is hoping you'll trip and drill a big financial hole in the ground, scaring off your investors. Gone are the days of a server in the kitchen auctioning Pez dispensers, that you can grow into Ebay, holding the first million off to the side, untouched. You have to think big AND small at the same time. How? Read more.

A rational business model can be the "kiss of death" to a start-up. It makes perfect sense, is perfectly executable, delivers on its promises, and is explainable in terms of team and plan. So why does it get a pass by investors?

Simple. Where do you take it? If it executes to plan, dollars flow in, and it connects with its market, we still can't get it into the big leagues, because the jump is too large. Scale matters.

You can do powerful internet startups on almost nothing these days - but that means there are too many with too long to get to the biggest leagues. And too many ways they can go wrong along the way. An "insanely great" Internet start-up must break the rules - well, at least some of them. If they break too many, they go the way of "dot bombs", or worse, WorldComm.

Take Facebook, which is considered a "nice small business", with a top value under a billion. But compared to many other Internet acquisitions, you'd think this to be big.

But not on the scale of a Google.

The Internet is becoming bimodal - a small handful of really big firms, and a very large number of much smaller firms. Smaller firms, like traditional small business, pay out the fewest pennies, and scrape in dollars, viewing business models in a conservative light. Big firms view business models in a disruptive way, ripping apart a rival's long held business by attacking its root assumptions - in the case of the Internet, substituting cheap bandwidth and distribution for more expensive paper, air time, and customer acquisition. You can also trade content and production costs for publicity. But for the big to be different than the small, the big have to have "insanely better" deals - for example, they need to be able to serve an hour length video (say 400 MB) for the cost that the small pays for a web page (say 20KB). A net advantage of 20,000!

Which is why you build your own datacenter as Google does.

In the "dot com" explosion in the 1999-2001 times, you used VC money to build a 3 Tier client server web facility inside a colocation facility, and negotiated a bandwidth deal for 3-10 megabit/s months, and that got you all the scalabity you needed from beta to IPO. I did this a few times enough to know.

But now its in the tens to hundreds of gigabits that big requires. How do you do this? Lets look at YouTube.

First, don't spend time on trying to monetize content, or even optimize your bandwidth usage - you simply don't have the pull to get all you need - you can't negotiate the deals, because you are starting as "small" and you'll only be "big" after you survive.

Don't worry about all the envy, for you'll always be put down no matter how good you do with what you have, because it will never be good enough. The point is to be better at gathering a community (and not just eyeballs), which one of the "big" can't really do as well as. This is the only thing that matters in the end - that "small" can deny "big" a top billing indefinitely.

As to copyright issues, well, that's a problem for the big acquirer post acquisition. Which is where Yahoo got cheap, trying to charge it off pre money to cheapen the deal. No surprise here - Google is beating Yahoo every time when Yahoo gets cheap.

And lets take a look at the copyright bogeyman. According to the Stevens decision:

• media needs to consist largely of copyrighted materials


• needs to have no way to remove


• needs to rely on piracy in the business model

Unlike Napster, YouTube content isn't dominated nearly 100% by exact copy pirated content, attempting to be a free source to create CD/DVD copies from. Lets look at a bigger picture before we strain for gnats, like mechanical royalties.

For Google, the click fraud issue is a far bigger one. The issue of collusion between Google and website traffic is far more serious, since both profit from the advertiser's misfortune. And its in the advertisers interest to assume fraud in the most common case - that the advertisement doesn't work. Could you imagine how the ad business might of faired if you could've proved such a link with print?

But in all Google's businesses, its best to realize that the bandwidth driven business model puts them at an financial advantage over competitors that goes a long way to funding a substantial legal defense. For Google is just out pennies of bandwidth for cost of goods, instead of dollars.

Also, the amount of material is overestimated. Those claiming this include satires, outtakes, short inserts, theme borrowings and other artistic uses covered by the doctrine of fair use. The biggest rated clips on the front page are usually totally amateur made videos. They don't need to compel piracy to have a legitimate business. However aggravated Jack Valenti gets about "Rip, mix, burn", and no matter how many Eldred cases Larry Lessig loses, its still a long way to tracking down some kids that supply humorous inserts to TV outtakes, and expecting to shake them down for a million or two.

There will be lawsuits. Google has as big a legal department as Microsoft does. But there have been many who have signed on. And more to follow. Why?

The RIAA experience in legal retribution didn't solve the business problem of the music industry, any more than the Digital Millenium bill did. Apple's iTunes did more to find a reasonable path out of this quagmire than all the rest - to start building media franchises that were functional in an internet age. Perhaps this is why Hollywood will choose a less litigious route this time.

We heard these same issues at the beginning of Open Source, and they all fizzled out. With 386BSD, we never got sued - yet envious rivals kept calling for it day after day, for 5 years. Call it for what it is - jealousy.

Yet YouTube could only be small for so long before being bought by "big" - in this case, "biggest". For it would have suffered the death of 1,000 cuts if it were to go the IPO route, and also been the perfect target for quick settle lawsuits, just at the time of needing to post 9 or so quarters of excellent K-1's to a touchy bunch on analysts, listening to every bit of ridicule and bad news.

So the answer is that you act like a standalone property "big" must have, pay the costs of being "big" ($1M a month!) for as long as it takes for "big" to be irritated that they aren't "big" in that sector, and you ignore all the bricks being thrown. Its also slightly helpful to have some kind of modest revenue model so that there's some proof that you'll be profitable - but face it, your actual business model is the "big" one - post acquisition. The one you can't credibly pull off in growing from "small" to "big". Which is why you must be acquired to complete "The Google Test". And why there is no "Yahoo Test".