BBC vs ISPs: Readers blame monkey, organ grinder escapes
Relief for Ofcom, Highfield
Andrew's Mailbag At last, something that will cheer up Ofcom chief Ed Richards and the BBC's Ashley Highfield.
You'll recall that on Wednesday, Highfield posted out a "Screw You" to the ISP business - refusing to help pay the spiraling costs of distributing the Beeb's iPlayer content. Don't rely on us to build a content network, he suggested.
And Highfield even threatened to name and shame ISPs who dared to manage the problem within their own networks. A black mark for an ISP from the Beeb would have the same effect as Roger Cook running down the street after the boss. It's no idle threat.
So Highfield's message was stark. You'll have to put your prices up, he ordered.
But Highfield will be relieved to discover that a fair number of Reg readers agree that ISPs are the villain of the piece - and rather than blaming the organ grinder, are lashing out at the monkey.
Now let's take a reality check. The BBC is committed to making sure its material is available on various "platforms". (Highfield himself has committed to investing in the internet as one of these platforms, so it's not as if some high-minded principle is at stake). And that's the reason the BBC paid BSkyB an annual fee of millions for the satellite broadcaster to carry BBC programming [*]. The generosity that sees the BBC throw money at a multi-billion dollar corporation won't be extended to ISPs.
Then there's the cosy agreement worked out between the organ grinders. This guarantees a huge annual profit for BT.
When it "designed" the market, Ofcom allowed BT to implement CBC, or capacity-based pricing. It was a cynical move in several ways. It ensured the broadband adoption figure ramped up quickly - which is what the politicians wanted.
But it ensured that ISPs couldn't satisfy increases in demand. Most ISPs are ADSL resellers who have to swallow CBC. Initially, monthly rental costs were high, but these were then lowered, and the CBC raised to compensate BT.
Which left ISPs with even less room for manoeuvre, and highly vulnerable to increases in demand for traffic. And the CBC, as we'll see below, is maintained at far higher levels than the market would allow, if Ofcom made BT release more capacity.
So when Highfield demanded ISPs absorb the cost of carrying the additional traffic he's generated, he was gambling that people would respond by beating up the ISP. That's natural, since the BBC runs nonstop consumer "campaigns" beating up ISPs. But it ignores the bigger picture, and ignores the stitch-up.
So who's to blame?
For all the constraints they work under, the ISPs have been their own worst enemy. Lets face it they've spent the last few years actively lying to their customers. I think most people can understand the concept of contention if explained, and they can understand the concept of traffic limits. What the ISPs have been doing to pretending to sell what they don't have - by pushing their 'up to' 8M lines as if the customers can expect anything like that, and by pushing 'unlimited' when the ISP knows damn well that unlimited has absolutely zero relevance to what they are actually providing.
As such, the ISPs have engineered customer expectations, and now those expectations a coming back and chewing their backsides off. What they pay to BT is irrelevant to that.
Irrelevant unless you get the BT bill at the end of the month.
The Other Steve agrees:
"Relying on the customer's failure to read the small print is not the basis for a digital content strategy."
How true, and yet surely it could also read :
"Relying on the customer's failure to read the small print is not the basis for AN INTERNET SERVICE PROVISION strategy."
Hoisted by his own petard there, methinks ;-)
And Wesley agrees -
I'm sure that most people would agree with me when I say that relying on the supplier's failure to print the small print is not the basis for a digital delivery strategy.
As I'm sure you've guessed, I'm thinking "Unlimited*" here.
In truth, the ISPs created this whole mess by trying to be greedy. Charging each other on transfer instead of capacity, they created a structure where they end up overcharging one another for something that as, in essence, a fixed cost. Once the capacity is there, it does not cost more if it is used. If they stopped advertising "unlimited" and then turning around to slap people who use the unlimited in an unlimited way, they would have a lot less trouble all around. Maybe if they started being honest with costumers, they could get out of this expensive mess they dug for themselves.
Now the truth is much more subtle and interesting, as one ISP tells us. If you depend on CBC, you're screwed:
I am exposed to the Capacity Based Charging model both from BT (via 3rd party wholesale outfit 186K) and from LLU wholesale from Tiscali so I can see some differences. Nevertheless to get traffic from to and from my ADSL clients costs me 10 times as much in cost of bandwidth as to send or receive the same packets to the rest of the world. That includes you BBC. So it costs me probably 20 times as much as it did the BBC to deliver their content. Do I blame the BBC? No I don't. Do I blame BT? you're damned right I do.
The major component of this bandwidth cost is fibre backhaul from the exchanges. If BT were to sell dark fibre pairs to the LLU players this price would fall though the floor and low contention ADSL would be available to all. But BT won't sell dark fibre because they fear that their ipstream model and C21 network could be instantly replicated with a vastly lower cost base.
Someone tell me why the annual rental of 1 GB WES/LES/BES circuits costs a fortune compared to 100Mbps variant (which isn't cheap). It is the same fibre, just a different transceiver. If OFCOM had made the same pricing model reflect BT's costs for installing and maintaining fibre as it did for copper pairs in 2000 when LLU started we wouldn't be in this mess now.
So, BT, you need to upgrade your exchange links. Oh, and while you are at it deliver your central pipes as Gigabit Ethernet, not this 655Mbps ATM stuff of yesteryear. Since no one is going to buy a central pipe of that size outside of a known data centre, you could charge them a few hundred pounds to install instead of the tens of thousands. You could ..but you won't will you? Not until someone makes you. The sad thing is you would end up making more money not less. All those little ISP's could start buying direct from you again. Didn't ADSL competing against your valuable leased line and ISDN business in 2000 show you anything? It was a good thing.
It is the price of fibre between BT exchanges that is the barrier to the UK internet services. I think Easynet was instrumental in getting a better deal with the BES service from BT but what it really showed was that BT will do anything not to sell dark fibre.
As soon as an LLU player could bring on another neighbouring exchange using a properly IP routed mesh topology with 100 times the bandwidth for the same cost as today you would see IP Multicast working. You would see packets for geographically close communications not being tunnelled half way across the country and back. You would have local resilience. ISP's might even run BGP in every exchange with other LLU isps. The UK internet wouldn't depend on a few data centres in London. Local data centres would pop up everywhere. People are used to Latencies of between 30 and 50 ms. But imagine if that was the worst you ever got and now your local VPN traffic got there and back in under 2ms and your London vpn head end was only 5ms round trip.
Ultimately LLU is completely held to ransom by BT determining the price of getting bandwidth to the racks in the exchanges. Unbundle the fibre and all the problems melt away.
So there's your villain. Or as Cris Page puts it -
I wonder if this BBC guy is living in the real world? Stupid question... of course not. His employer is funded by the annual extortion practiced by HMG as an annual ownership tax.. His employer knows little of the really harsh realities of the real world
The BBC struck the BSkyB deal in 1998; it ended in 2003. Thanks to readers for the link.
And here's an argument supporting tiered pricing.
ISPs are in a weak position to criticize the BBC's idiotic tiered-service pricing proposals, because their existing pricing system is just as demented, and is fraught with perverse incentives that have put them in their present bind.
The thing is, broadband ISPs marketing materials currently pretend that they guarantee something like 1-2 MB/s for a flat rate to all their customers all the time, despite the fact that they know that their available total bandwidth is much less than that, and that if all their customers were to ship packets 24/7, they would all be getting the equivalent bandwidth of a dial-up 56k modem, at best.
This is really stupid, and is the real basis for ISP's freak-out behavior (and misbehavior) upon being confronted with the rise of P2P applications, which bring large increases in data transfer efficiency at the cost of customer computers saturating their supposedly available bandwidth 24/7. In this sense, Comcast's recent loopy public posture can be traced to a variant of the "Tragedy Of The Commons," except that it's more like a black comedy instead.
The solution, which nobody appears to be talking about, is not to make technical fixes to bandwidth-throttling router algorithms. What the ISPs should be doing is charging people for the real cost of their network usage. They should drop the hollow pretense of flat-rate bandwidth guarantees, and instead charge people a flat rate *per bandwidth-hour*. People who transact 2Gb of packets per month should be paying roughly twice what people who only do 1Gb per month, and so on.
A very rough US-centric calculation, assuming a typical 2Mb/s for $20/month current flat-rate contract for an ISP that expects you to use about one per cent of that bandwidth means that they value bandwidth at about $2.60/Gb. These numbers are probably a little squishy, so call it $2/Gb. That's what they *should* charge. Obviously the rate could be adjusted for off-peak-hour use, and so on.
Presto. No more perverse incentives. People start metering their own computer communications costs the way they do their phone and electricity, so the ISPs don't have to do it for them. The free-rider problem goes away. People who really want to watch TV on their computers bear the real costs of doing so.
Another benefit is that people might get a little more careful about their computer security, if they get charged for their commandeered computer whiling away its evenings by sending spam.
To some extent the free-rider problem now gets shifted to the P2P networks instead, since they now must rely on good-citizen community participation by network participants when participation actually costs participants something. Some P2P models might not survive the change. But those would be the ones that thrived due to a distorted bandwidth market. Presumably it would be possible to configure a P2P service that adjusts its costs and benefits to the real economics of bandwidth, so that people can afford to use it without going bankrupt. For example, I should be able to configure my P2P app with a monthly bandwidth-hour cap, with higher caps resulting in more available service to me.
Anyway, the larger point is, the BBC shouldn't be castigated for telling the ISPs how to charge their customers, at least until the ISPs stop doing such an incompetent job of it themselves.
Carlo Carlo Graziani
Highfield also raised eyebrows with his assertion that "The best technical solution is usually Moore’s law". An oddly ignorant thing to say, since the capacity and price of copper and fibre connections have very little to do with the density of transistors on a semiconductor die.
Finally, a bit of a cuffing for my making fun of Ashley Highfield's assertion that Moore's Law will solve the problem.
The man from the Beeb had a point. Silicon getting faster and cheaper means interwebs get faster and cheaper too, because at the end of the day the big long wire or fibre or chunk of spectrum is the expensive part; there are a *lot* of communications theory people working on maths to make lots of data fit where only a little data used to fit.Chip lasersDWDMsingle-mode fibreoptical switchingdoped fibre amplifiers
All are products of the semiconductor industry.
I would have thought it has everything to do with the density of transistors on a semiconductor die (at least for fibre). The kit you terminate fibre with is where nearly all evolution of networking capabilities lies, and they are largely bound by semiconductor technology. And considering we're largely talking about backhaul, as opposed to local loop capabilities, fibre is all that's really that relevant.
If you know different, do share!
Fair enough. But surely, if the price of bandwidth is artificially kept artifically high by BT by keeping dark fibre out of the market, then the cost and capability of semiconducters pales into insignificance?
Any readers care to break this down for us (and Mr Highfield)?
Finally, here's something I never knew before:
You're quite right when you say "Ceci n'est pas un pipe" (it certainly isn't) but I've a feeling you meant to say "Ceci n'est pas un tuyau". Whilst Mr Highfield did show ignorance with his comment applying Moore's law to bandwidth, I'm pretty sure he's aware that it has nothing to do with a blow-job ("un pipe" en francais)!
Keep up the good work (but perhaps let someone else check the translations), Alasdair
Then there's obligatory no-comment comment:
BBC tells ISPs to get stuffed - wot no comments
Fix it please
Thanks for sharing, mdl. I hope the last two pages explain why. ®