Jerry Brito

How to include an entire web page in a URL, no server needed

This is a pretty cool little nugget I picked up from the excellent Security Now podcast and wanted to pass it along. Basically, there is an old and obscure IETF RFC that standardizes a scheme for including data in a URI, and all modern browsers seem to support it. In English, this means that instead of using “http:” and pointing to a web page hosted on a server somewhere, you can use “data:” and include the web page right in the URI itself!

For example, if you click on this link, you’ll see another version of this post, but it’s not hosted anywhere. It’s completely in the URL itself. I made it using DataURL.net, which lets you create “data:” compatible base-64 encoded URIs by simply dragging and dropping a web page, image, or other file.

Now, the reason this was being discussed on Security Now is that some smart black hats out there have been using the scheme to create phishing links. There are, of course, legitimate uses. The most often mentioned is speeding up web page load time by cutting down on http requests. But it seems to me that there have got to be other unique and interesting uses.

One that springs to mind is a simple alternative to TwitPic or Twitter’s hosted photos. That is, share photos without hosting them on someone else’s server (and becoming subject to their terms and conditions). To set up a service like this one would have to not only offer encoding like DataURL.net, but also link shortening of incredibly long URLs. I’m going to noodle with this idea a little more and maybe see if I’m up to the task. What other uses for DATA URIs can you think of?

How Google Fiber is not like Verizon FiOS

In a recent post, Tim Lee does a good job of explaining why facilities-based competition in broadband is difficult. He writes,

As Verizon is discovering with its FiOS project, it’s much harder to turn a profit installing the second local loop; both because fewer than 50 percent of customers are likely to take the service, and because competition pushes down margins. And it’s almost impossible to turn a profit providing a third local loop, because fewer than a third of customers are likely to sign up, and even more competition means even thinner margins.

Tim thus concludes that

the kind of “facilities-based” competition we’re seeing in Kansas City, in which companies build redundant networks that will sit idle most of the time, is extremely wasteful. In a market where every household has n broadband options (each with its own fiber network), only 1/n local loops will be in use at any given time. The larger n is, the more resources are wasted on redundant infrastructure.

I don’t understand that conclusion. You would imagine that redundant infrastructure would be built only if it is profitable to its builder. Tim is right we probably should not expect more than a few competitors, but I don’t see how more than one pipe is necessarily wasteful. If laying down a second set of pipes is profitable, shouldn’t we welcome the competition? The question is whether that second pipe is profitable without government subsidy.

That brings me to a larger point: I think what Tim is missing is what makes Google Fiber so unique. Tim is assuming that all competitors in broadband will make their profits from the subscription fees they collect from subscribers. As we all know, that’s not how Google tends to operate. Google’s primary business model is advertising, and that’s likely from where they expect their return to come. One of Google Fiber’s price points is free, so we might expect greater adoption of the service. That’s disruptive innovation that could sustainably increase competition and bring down prices for consumers—without a government subsidy.

Kansas City sadly gave Google all sorts of subsidies, like free power and rackspace for its servers as Tim has pointed out, but it also cut serious red tape. For example, there is no build-out requirement for Google Fiber, a fact now bemoaned by digital divide activists. Such requirements, I would argue, are the true cause of the unused and wasteful overbuilding that Tim laments.

So what matters more? The in-kind subsidies or the freedom to build only where it’s profitable? I think that’s the empirical question we’re really arguing about. It’s not a forgone conclusion of broadband economics that there can be only one. And do we want to limit competition in part of a municipality in order to achieve equity for the whole? That’s another question over which “original recipe” and bleeding-heart libertarians may have a difference of opinion.