Back in 1965 Gordon Moore (Intel Co-Founder) made the empirical observation that transistor density on integrated circuits doubled nearly every 18-24 months, later becoming known as Moore’s Law. More transistors translated into more processing power. More processing power meant beefier programs and, unfortunately, sometimes lazier programmers.
System Requirements? I Don’t Need No Stinking System Requirements
The growth in processor speed used to be noticeable. As a kid in the 80’s and 90’s I remember frequently upgrading from one processor to another so I could take advantage of the current killer app. Now I don’t even read system requirements anymore. These days the only practical programming constraint for the desktop is the theoretical processing time (whether the algorithm’s job finishes in a fixed amount of time or has a tougher job that multiplies its work exponentially).
The only consumers that care about processors any more are gamers. Growth in processor speed is chugging along nicely behind the scenes, but everyone else has moved on to the internet because networked applications are more engaging. Consequently, Internet bandwidth is the new bottleneck for interesting applications on the computer, and it has been for a while.
Nielsen’s Law of Internet Bandwidth
Jakob Nielsen (the creator of a lot of UI advice severely underrated by CS students) made this point back in 1998 when he wrote about Nielsen’s law of Internet Bandwidth. Basically he stated that bandwidth grows by 50% every year. See chart above.
Right now I’m in a cafe on a 4 - 6 mbps connection according to Speakeasy. That matches up pretty close with the Neilsen estimates starting with 1000 bps in 1986 and ending with about 5 mbps in 2007. Even with this growth, however, it’s still frustratingly slow. I’m still tempted to chuck my wireless mouse at something when a page doesn’t load.
Fast But It Could Be Faster
IT DOESN’T HAVE TO BE THIS WAY. The US actually lags behind other OECD countries in broadband speeds (even through we have high broadband penetration, 62% or something, that’s based on a laughable 200 kbps definition). Japan’s surfers connect to the internet on a 100 Mbps Ferrari to our 1970’s 40 Mbps hatchback (Not a Truck Senator Stevens). They also pay much less, $0.22/Mbps to our $3.10/Mbps. And to think companies brag about a $260/month 50 Mbps connection in Sacramento.
Of course this has nothing to do with big telco’s near dominance of broadband. Nothing to see here, go back to what you were doing. Hey, well that’s what we get for having a congress that call the web a “series of tubes“.
So What? If (slow) { BuyFasterInternetz }, Right?
The problem with this limitation on computing is that it can’t be solved all on your lonesome. The net is a network, which means we all have to pitch in to make things faster. Websites won’t design more complex sites unless users get faster connections. Users won’t get faster connections unless sites require it. Neilsen, that perceptive bastard, noticed this before and recommended sites design behind the curve to reach the most users.
Another problem is that websites are moving away from just being a relationship between one user and one site. Services are fragmenting into specialties which means a lot of sites are serving up third party content along with their own, furthering the dependencies. Most commonly, these third party services are widgets.
I Can Haz Faster Internetz?
Well, so far the internet has been sucking at being fast, but I have a lot of faith in things changing in the near future. Why? Video (YouTube) is driving demand for higher bandwidth and telcos are working hard at getting it to people (well not that hard). The expansion of wireless networks allow carriers to not only penetrate markets with superior voice coverage, their bread and butter, but also super awesome broadband wireless (see sprint and verizon).
Unfortunately once again the fox is guarding the hen house, and the big telco’s, like all oligopolists, will ride the high prices at the top of the demand curve for as long as possible (till politicians and the FCC stop taking their money). Google and the 700 Mhz spectrum just might be enough to put a little hustle back in them.
