Just above we have a posting from an ISP who says he’ll gladly host a Google cache for free. In fact, I’ll bet that Google will be able to place an edge cache at the site of any ISP it wants. Probably for free, because Google is big. Google, YouTube, and its related services consume SO much bandwidth that no ISP would say no. The ISP would easily save big on its backbone connection. Terabytes per month on YouTube alone.

But would the ISP afford to allow just anyone to get free hosting by putting a cache at their sites? Doubtful. Caches take up space and power and require access for maintenance. If a small startup were to call your local cable company and ask for “co-location space,” the person there would probably say, “That’s not a product we sell to the public.” That is, if the person who answered the phone even knew what it was.

And of course, would-be competitors won’t be able to buy space on Google’s private edge caches.

So, in what way is this neutral? Google can get its servers into places where CoolNewInternetGarageStartup.com can’t.

So, Google is still getting preferential access to infrastructure It’s just co-location space instead of pipes. And this is even a more difficult thing to buy than bandwidth. Any Internet carrier will sell you a pipe. But co-location space at ISPs, which is more cost-effective than buying pipes, isn’t necessarily even for sale to you unless you’re Google. So this is really, really non-neutral and anticompetitive.

@Paul: Structural separation, meaning the ban on ISPs selling content delivered over their networks, is the primary demand of Isenberg and Obama transition team member Susan Crawford. They want ISPs to be confined to operating the network, and banned from collecting money for any content; they’re supposed to be happy with collecting access fees from retail and wholesale customers.

In effect if not in name, caching inside the ISP network improves reliability and quality of content delivery, so it is a means of effectively boosting priority. Granted, it’s an end-to-end system, but when the end points are only a thousand feet apart, you’re in a whole different ball game than when they’re a thousand miles apart; the network becomes a LAN rather than a WAN, and standard congestion rules don’t apply.

If such deals were exclusive, regulators would smell a rat instantly, but making them non-exclusive but only practical for a hand full of players hides their nature in the fog of regulatory uncertainty.

It may be hair-splitting but I can see a huge difference between, say, A) a service provider who cuts an exclusive deal to provide video from one content provider over its broadband link to the customer and slows, blocks or excludes (by not allowing similar deals) other content providers; and B) a service provider who cuts deals with content providers of all stripes to give the customer the best access to the content of their choice. The former seems like something that should be regulated (because consumers without adequate choices of broadband providers have no economical power to leave for another provider), while the latter seems like good business for everyone involved — consumer, content provider and service provider.

No matter how fast Google makes its connection to the ISP, it doesn’t gain anything if the end user doesn’t click there; and unless Google is paying for exclusive connections that exclude end users from choosing other sites, how is it hypocritical to their stated net neutrality principles?

The third pipe is growing — and not just 3G and 4G, which are mobile technologies whose speed is intrinsically limited by the need for mobility. WISPs like my own are actually increasing in number. If the “net neuts” don’t get legislation in place to squash us, we will likely be THE third pipe.

One quibble about a point above, however: P4P is absolutely not ISP-friendly. Especially to small ISPs, whom it puts at a disadvantage. (Larger ISPs are more likely to have one or more users who are offering a file that another user wants, so they may save on bandwidth as a result of P4P. But this will virtually never happen on a small ISP’s network.) Thus, P4P is essentially an anticompetitive tool that can and will be used to try to impose a duopoly and drive small and independent ISPs out of business. It’s also very bad for rural providers, whose bandwidth is exceedingly expensive.