WINDING DOWN
An idiosyncratic look at, and comment on, the week's net and technology news
by Alan Lenton
As promised we're back! Just in time to celebrate Winding Down's Ninth birthday. Yep, we've been turning out the newsletter for nine years now. The question is, will we make it to be ten? I hope so!
In the last issue before the break I promised to have an extended look at the so-called 'net neutrality' issue that's causing so much anguish in the US at the moment. I'm happy to say that I finally managed to come up with a piece, which takes up all of this issue. As a disclaimer, I'd point out that I'm not neutral on this issue, but I don't like any of the solutions proposed. A bit of a cop out, you may suggest, but then I never claimed to be able to solve all the problems of the Internet, the world, and the universe by myself - even if I do know the answer (it's 42). The tech wizards on the Nanog mailing list are working on the first, Bill Gates is working on the second, and the third I'll leave to Steven Hawking.
What I can do is to try and provide some help on what its all about - so, here is my take:
Analysis: The Net Neutrality 'Debate'
A month or so ago a joint policy proposal from Google and Verizon on net neutrality opened a Pandora's Box that's been sitting around for a long time. For many, the fact that the authors were among the biggest players on the Internet immediately tainted the proposal's content, for others the proposal was a welcome breath of fresh air.
The blog-o-swamp seethed with opinion, many based more on who the proposers were than on the content, and the level of vitriol of some of the contributions rivaled that seen in the flame wars of the early Internet.
So what is it really about?
To understand the issue you need to be aware of three things: how the Internet works, what the technical problems facing providers of the different services and delivery mechanisms are, and what the business problems facing ISPs are. Let's look at these in turn.
The Internet uses a protocol called, unsurprisingly, Internet Protocol (IP), to move traffic around the net. From our point of view the most interesting thing about the current generation of the protocol is that it is a 'best effort' packet protocol. It splits up the data into packets and then sends them around the network via the fastest route it can find for each packet. At the receiving end the packets are re-assembled. It's very efficient from a network usage point of view, but it makes no guarantees about how long it will take for each packet to reach its destination.
That can be a problem for some types of application, but not for others. For instance, voice conversations need the packets to arrive in sequence to have a decent conversation, but, in most Skype conferencing I've been involved with, 90% of packets are carrying the audio phrase, "Are you still there?" A pretty clear indication that the packets are being subject to delay. On the other hand, it doesn't matter if it takes packets carrying email data a little longer to arrive - it's not a real-time application. IP is neutral in the sense that it treats all packets the same, and doesn't prioritize on content type. (For more about different traffic problems, see my article.
The technical problems facing providers of services on the net are concerned with optimizing the speed at which their customers are able to access the services. Hence Google's interest, since it owns YouTube, and it's not helpful for the user experience when its video packets get stuck behind a bunch of spam packets...
The ISPs suffer from a different technical problem, peak load capacity. Few providers actually have the capacity to cope with all the traffic passing through their networks at peak times, so they already indulge in what is euphemistically known as 'traffic shaping', that is prioritizing some traffic over others in an effort to give all users the best possible experience.
Regular analog telephone networks have always suffered from this peak time problem, but their solution doesn't have the same psychological consequences as network congestion. If a phone system is overloaded you get a busy signal, and you can't make the call. If a network is overloaded, you still get a 'connection' but it's slow and erratic, speeding up a little just as you are about to give up, and then slowing down again when you decide to wait. The latter is much more frustrating than the former. The ISP techies do their best to speed things up for you, but what they can get out of the current system is often limited.
Finally we have the ISPs' business problems. To all intents and purposes the basic business of ISPs is that of a public utility. This means that they suffer from the advantages and disadvantages of utilities. Everyone needs the product, which requires infrastructure to deliver, and so is a natural monopoly. This means a guaranteed profit, usually regulated. The problem is that as a utility it's difficult to grow and provide higher dividends for their shareholders. This is why some of the biggest failures in recent years have been utilities dabbling in dodgy, or even fraudulent, other activities (think Enron).
In the USA, in most areas there is now a broadband duopoly - usually AT&T or Comcast - and it shows in the prices. With two exceptions consumer prices in the computer/networking business have dropped dramatically over the last ten years. The exceptions are the Windows Operating system, where Microsoft holds a near monopoly on desktop and laptop operating systems, and broadband network subscriptions. The broadband infrastructure in use now was put in earlier in the decade, and in that time usage has gone up from 20% of the population to 65% of the population. Adding a subscriber costs virtually nothing since the original capital costs have long since been met.
Consumer broadband is lucrative, but limited in the sense that there are only so many potential customers, and nearly all of those who want broadband now have it. So... The ISPs are starting to look at how they can change the way the Internet works to provide new revenue streams. There are a number of possibilities. They could charge the people supplying data to have their packets given priority over other people's packets. They could prioritize their own other services, like Internet TV, over their competitors. They could provide, at extra cost, some sort of 'gated' service that keeps out the riff-raff on the public Internet. All these, and more, sophisticated variants break the fundamental idea of the net that all data packets should be treated equally. And they are all a way of charging more for something they already provide.
The Verizon/Google proposal came up with a slightly different way of breaking net neutrality, which on the face of it seems to be a little more fair that those solutions so far proposed. Essentially, their proposal is that the packets on the Internet should be prioritized on the basis of the type of content. This means that rather than Google paying the ISP a packet (so to speak) to have all its data given priority over everyone else's, both YouTube data and your streaming video to grandma showing the kids' birthday party would travel at the same priority, which would be higher than that of packets carrying text, for instance.
This sounds pretty reasonable, and I'm sure that most people would agree that this is a good idea. Unfortunately, this is where the law of unintended consequences cuts in, and a little more consideration of the idea reveals that it's not quite a good as it seems, and has a number of potential problems.
The first is, how do you tell whether a packet is carrying video or not (taking video as an example)? You can't rely on the sender to tell you. Everyone thinks their traffic is the most important and will therefore cheat. This means the dreaded 'deep packet inspection' - i.e. the ISP will look at the contents of every packet to ascertain what its priority should be. Not a good thing for the privacy of your communications, business or personal. And what about encrypted material? How do you tell what that's carrying? The truth is that any halfway competent network programmer could make all the output from his or her program look like video, if video had the highest priority.
Secondly, who is to say what has priority, and what its level should be? Should Flash content have the same priority as streamed video? What about the new HTML 5 - how do you assess what the impact of new stuff will be? Priority would undoubtedly be given to existing material and new innovation would, in a manner of speaking, die a slow death!
So, is there a solution? I don't know; nothing I've seen so far even approaches something resembling a solution to the problem, short of a massive optical fibre build out that provides unlimited light speed access to everyone, and that's not very likely, this side of the heat death of the universe. What I do know is that there is no silver bullet. Obviously the problem will be solved, one way or another, but the compromises that will have to be made will be painful for both sides of the debate, and satisfy no one. Hopefully, though, this piece will allow non-techie users to make their own evaluation of the new proposals that will come out of what's already been put on the table.
And the only thing remaining in Pandora's Box is hope...
No URLs this week, just Google for 'net neutrality' and you'll get more than you ever wanted to know...
Acknowledgements
Thanks to readers Barb, Fi, and to Slashdot's daily newsletter for drawing my attention to material used in this issue.
Please send suggestions for stories to alan@ibgames.com and include the words Winding Down in the subject line, unless you want your deathless prose gobbled up by my voracious Spamato spam filter...
Alan Lenton
alan@ibgames.com
19 September, 2010
Alan Lenton is an on-line games designer, programmer and sociologist, the order of which depends on what he is currently working on! His web site is at http://www.ibgames.net/alan.
Past issues of Winding Down can be found at http://www.ibgames.net/alan/winding/index.html.
|
