The critical assumption behind this article is that the ISPs "slow lane" - i.e., the general internet - will degrade to the point where it isn't usable.

Now, the FCC claims to be planning regulation to prevent this, but it's unsurprising that people don't trust them.

However: the "slow lane" is still going to be most of the internet. The question becomes, will enough of a typical ISPs customers use *only* those mainstream, big business web sites able to pay the ISP's bribes (and assuming that they are willing to do so) that it is feasible for the ISP to lose the rest?

I find it doubtful, but if anyone has statistics it would be interesting ...

"I pay my ISP for a certain bandwith. They should provision for that." ... there aren't enough people willing to pay the amount this would cost to make such a business feasible. The internet works, and is more or less affordable, precisely because you share the backbone capacity with your neighbors.

"Comcast HAS implemented caps" ... now, that's very odd. Why would anyone join Netflix in the first place if they've got a connection with data caps? Doesn't it make it pretty much useless? You'd burn through the cap in no time ... we run out of data (or nearly so) most months, and that's just from Youtube.

I don't know what makes you think I've got a particularly strong opinion - oh, on the general principle behind the FCC's reformulation of net neutrality, sure, but the details of this specific incident aren't directly relevant to it.

Sounds like you're trying to change the internet to a COD model, where you pay for traffic received rather than traffic sent. I'm not sure that's realistic. (For one thing, it would make DDoS attacks even more painful than they already are!)

Also note that Netflix customers using Comcast really aren't already paying for those bytes. To do that, Comcast would have to identify Netflix customers and charge them extra, and you can just imagine the howls that would cause. Oh, sure, they could charge by the gigabyte or implement data caps. But I don't think their customers would like those options either. Or they could just up their prices across the board, but then Netflix users would be being subsidized by everyone else, and I don't think that's fair. The best solution, IMO, is for Netflix to pay and pass the cost on to their customers, and that's exactly what's happened. AFAIK, there's nothing stopping them from charging Comcast users extra to cover it.

(In the thread you linked, it didn't sound like the OP was talking about an ad-hoc VPN to me; that also means that his office connection probably wasn't via Comcast, if Comcast don't do enterprise. Of course, that's all just speculation. OTOH, I still figure Netflix would have sued if they'd found any actual evidence of discriminatory throttling, which shouldn't have been hard to do.)

"Different routing inside their network [...]" ... actually I was thinking more of the backbone routing. They've got a big network, presumably they have multiple backbone exit points, and it seems entirely plausible that some have better connectivity to Netflix than others.

"none of the traceroutes I have done suggest that a business account is routed differently" ... odd. You wouldn't expect a business account to be competing with residential accounts for bandwidth. I suppose they could be using QoS over the same routes, though. (Or just cheating their business customers. Take your pick.) ... or perhaps we aren't even talking about the same thing? I'm talking about enterprise-level connections, for businesses with hundreds or thousands of machines, usually including servers. If you just meant small retail or home office connections, that's a different story. (But small retail and home office don't usually have VPN, so I don't think that's likely to be the situation described in the forum thread you linked to.)

"'peering' with someone who doesn't own a network is just bizarre" ... well, that's just nomenclature. Call it what you like, the end result is the same.

I still think the bottom line is that Netflix was trying to avoid paying their fair share of the costs. Did you see the part where they threatened to generate junk traffic from customer's machines, specifically in order to incur extra costs for the ISPs? That doesn't sound like someone negotiating in good faith to me.

Actually, even if true, it still doesn't prove anything unless you can show that Comcast routes their business traffic in the same way as their residential traffic, which seems unlikely.

Of course, to turn it around, I can't prove that Comcast *weren't* intentionally sabotaging Netflix. But if they were, then the changes in the FCC's position (at least those described here) aren't relevant; Comcast's alleged behaviour would presumably be just as illegal under the new proposals as under the old ones.

"The home connection (with the slowdown) and the office connection (without) hung off of the same router deep inside Comcast's network."

Huh? Where did you get that information?

So, if you change the traffic route so as to bypass the bottleneck, you get better service? That doesn't prove anything about the nature of the bottleneck.

It doesn't make any difference in the long run whether the specific equipment involved is already present or not. Any additional load on that equipment brings forward the day when it needs to be replaced. Bottom line, why should Netflix get a free ride at Comcast's expense?

My understanding is that Netflix paid for direct peering with Comcast, in order to bypass Netflix's backbone provider. I would expect that to involve actual equipment, though I don't claim to be an expert.

Still, even if all they had to do was ask their backbone provider for more capacity, surely said provider is going to charge for it? What would be the point of under-utilizing a link if full utilization didn't cost anything?

Of course there isn't "special equipment", don't be ridiculous. I said *extra* equipment, to carry the bandwidth Netflix wants but didn't want to pay for.

No ISP ever promised their customers a congestion-free internet. Like most people, I use only a fraction of my bandwidth on average. Why should other people pay extra for a level of service they neither need nor want, just to subsidize Netflix's customers?

I believe it does.

Go ahead; no doubt you can arrange for a business-level connection if you're determined. You'll have to pay at least ten times as much as you do now, but that's your choice.

Or you could tell Netflix to sort it out - which they've now done, of course.

It seems rather unlikely that infrastructure upgrades planned for 1996-2006 would have been expected to provide the level of service we now take for granted, never mind the level of service Netflix needs. Do you have figures?

I disagree. Netflix created the problem. Everything was fine when Netflix were with their previous ISP.