Google’s cost savings on hardware have historically given them a competitive advantage.  In their early days they stuck to budget price/performance hardware machines, which compared to the high-end servers their competitors used gave them a 10X cost advantage in terms of CPU cycles per search.  Sometimes a programmer will invent a new algorithm that improves search relevance, but if the cost to deploy the algorithm is too high, the algorithm can’t be used.  Google could afford to spend more machine CPU for each user and this helped them give users better results.

Their storage costs were cheap too, which let Gmail offer far more storage space than their competitors inititally.

Now, with hundreds of thousands of servers powering their service, there is a huge capital barrier to entry for would-be competitors.  A few programmers won’t be able to out-Google Google with a hodge-podge of PCs anymore.  The costs of entry into the search space have gone way up.

If Google is assembling their own worldwide network backbone, maybe aside from any possible consumer applications such as launching their own ISP, they’ve simply decided that using this resource is key to their success, that there are cost and strategic reasons they want to control it.

Crawling the net is expensive...  Consider that every n days Google essentially copies the entire web onto their servers.  The entire web needs to move over network pipes into the Google datacenters.  If they want to have a fresher index, they need to crawl more often.  This is a lot of bandwidth…cheaper bandwidth means they can have a fresher index for less money.  What if Google’s crawl costs were 1/10^th their competitors?

My study "The Google Legacy" touches on this topic. Three key points: [a] most people when learning about the scope of the existing Google engineering can't grasp what Google has built, is operating, and scaling; [b] experts may see speed as a function of money, not an outcome of hardware and software engineering; and [c] finance-oriented individuals understand the bottomline benefit Google's operating efficiencies deliver yet lack hard data to put these numbers in context.

What's the result? Two examples fall readily to hand. First, at least one Yahoo engineer at the Yale forum is convinced that Google can be matched and beaten. The only hitch in the git along is that Yahoo is not a homogeneous computing platform. There are some cost issues ahead for Yahoo as it makes an effort to normalize AND deliver new services. Yahoo has grown by acquisitions. Google buys companies and shoehorns their functions into Google data centers. Elegant, no. Efficient, better than creating a fruit salad of technology. Second, Microsoft has billions of dollars and legions of super-wizards. The needed job of getting Windows Advanced Server to tiptoe around the message traffic bottlenecks AND keep the legacy agenda is more than a hard day's work. In fact, turbocharging clusters of servers running Advanced Server is a reasonably big job. So, in these two examples, both Yahoo and Microsoft are in the process of addressing their different system issues. I know that both will catch up -- at some point in the future. The key point, of course, is that Google's comparatively homogeneous system has good speed (2,000 megabyte per second reads, according to one engineer familiar with Google benchmarking), will benefit from bigger pipes (the Gildereseque dark fiber metaphor) among its data centers, is up and running, and undergoing incremental enhancements (I watch my GMail storage click upwards each time I check my Google Mail). When the number two (Yahoo) and number three (MSN) catch up, number one will probably be keeping a reasonable technical lead. More of this antithetical thinking appears in my study. See www.infonortics.com for a table of contents and a sample chapter.

S. Arnold from Harrod's Creek, Kentucky, 7 45 am Eastern

When I first saw “network” as the topic, my take was more about the value in the digital assets of the network rather than the physical assets.

IMHO: It’s the Platform, Stupid.

What really creates sustainable value is how rich the platform (search technology, content & data, applications & tools) is, and how vibrant the ecosystem (end-users, content publishers, developers, channel partners) is in which everyone takes part.

However, Google’s DNA of secrecy could be their Achilles heel. Google is churning out a lot of great (and not so great) technology to build their platform, but so far they’ve been missing the boat on telling people what the vision looks like – and in the process, they’ve been doing a better job at scaring off prospective channel partners than doing platform evangelism.

So while Google is trying to build out proprietary physical network assets, and Microsoft plays catch up shifting their desktop assets to the web, Yahoo actually has a decent shot at catching both of them by surprise – they’re the only company really delivering on Web 2.0 evangelism right now.

Ultimately, no matter how talented & smart Google is, no matter how cool their products are, they will still have to convince other developers & partners their platform is a place where everyone can win… not just the people inside the Googleplex.

-Gautam Godhwani
www.SimplyHired.com

Having hatched a plan to download the world's libraries and begun a research effort aimed at cataloging people's genes, Mr. Brin hungers, with the boundless appetite of a man who has obtained great success at a tender age, for the one place Google has yet to directly penetrate - your mind. "Why not improve the brain?" he muses. "Perhaps in the future, we can attach a little version of Google that you just plug into your brain."