More on net neutrality

The latest stratechery (now pronounced "strah-tek-ery", apparently) podcast had a lot of good information in it.  There was discussion of what "common carrier" status means, which I should have probably explained better in my last mention of the issue.  And there was an excellent discussion of what "rent seeking" means, in economic terms.

(And a brief sidenote about the Andreeson quote.  Utilities probably did have 10x as much poop to handle every three years for the first decade or more of laying pipes to deal with that.  I wish my grandfather was still around to ask about that; he spent most of his career putting in sewer lines.)

There's also some interesting things I never considered, like how Netflix' behavior might be rent-seeking, and how metered usage would solve a number of problems.

The problem with metered usage is a simple matter of cognitive load.  Until the prices get much lower (in line with the cost of providing the data), it adds much great transaction costs.  Very few people would have signed up for broadband at all, if it had been metered; they had no idea how much they used.  And even now, it'd be a big shift.  But maybe that's where things need to go; I've been solidly against that, but they've got me thinking a bit more about it, and maybe it's inevitable.

But there were a couple places where I thought the discussion fell down.  One was that they failed to mention how, back in the 90's, the telecomms were given hundreds of billions (yes, that's with a B) of dollars in tax breaks to provide Fiber to the Home (FTTH).  How much of that exists?  Yes, basically zero.

Another was that there was no mention of how the governments provide rights of way to put in cables.  That's also a hundreds of billions of dollars subsidy.  Not that I mind it being provided; I do agree with them that, at this point, internet access is a basic right.  But it does bear on the discussion.

A final one (and this is the important one) is the idea of content delivery networks.  The basic idea is that the CDN provider (Akamai, for instance) puts their servers in the ISP's data center, and Netflix, for instance, provides data to the CDN, and the CDN delivers to the customers.

What that means is that, despite the fact that Netflix is the bulk of all internet traffic, most (read: probably 99% or more) of their traffic is only going from CDN to end user.  And that means, as far as the ISP is concerned, the data is staying within the ISP.  And bandwidth within the ISP is, effectively, infinite.  The cost to the ISP is the cost of electricity.  In aggregate, that's a lot, but it's mostly the cost of having the system operating; data traveling adds very little.

So I fundamentally disagree about Netflix being rent-seeking.  And about the reasonability of the ISP charging extra for Netflix.  Netflix pays the CDN, and the CDN pays the ISP, so they're actually looking to triple-charge, not double-charge.

The other major issue is that they missed, as I mentioned, Wheeler mentioning "common carrier" status, and mentioning reclassifying ISPs.  Not a huge deal, as I don't believe Wheeler when he talks about it, but it's worth mentioning that Wheeler did add it to the discussion.

Still, well worth a listen, despite it being quite long.

No comments:

Post a Comment