Yesterday, as I’m sure many of you are aware, the European Parliament voted to reject the ‘net neutrality’ fundamentalist amendments to the already flawed proposals they had helped to create. That’s the good news. The bad news is that the law that we now have is merely ludicrous, rather than insane. Furthermore, it doesn’t properly protect end users, hold ISP feet to the service delivery fire, or truly encourage broadband ecosystem innovation.

An inevitable collapse of a bad idea

The core conceit and fallacy of ‘neutrality’ has been to create a doctrine of fairness to packets at the local mechanism level. I hate to have to tell you this, but packets don’t have feelings. They don’t get irate when they are in a queue too long and other packets jump ahead of them.

Hundreds of pages of books and papers have been written about differential traffic management and ‘discrimination’ and ‘congestion’. These phenomena are never quantified, so their objective nature is merely assumed. In reality, the words on the paper are technical nonsense, in the strict sense of ‘nonsense’.

The pile of literature on ‘net neutrality’ has been a waste of human effort and a loss of good wood pulp. The only thing that matters is the service we deliver to end users, not the fetishized mechanisms of the network. The user experience is a product of all those internals interacting via a statistical process, and is an emergent phenomenon.

The packet pixies are capricious

This is a very profound fact: all of the books and papers out there that I have seen fail to mention this or grasp its implications. Try searching for ‘stochastic’, ‘statistical multiplexing’, ‘nondeterministic’, ‘randomness’, etc. and you will soon realise these aren’t books about any real broadband network ever deployed! If it wasn’t so tragic and serious, it would be raucously funny.

Any particular outcome (good or bad) for any particular user of any particular application is merely an act of chance in the network casino. That you ‘won’ for the last week or decade, doesn’t tell you that the casino exists to fund your lifestyle. That you suddenly ‘lost’, doesn’t mean the casino has become a naughty ‘discriminator’ robbing you of your dues.

Don’t just take it from me. As the Broadband Internet Technical Advisory Group says [PDF], “The absence of differentiation does not imply comparable behavior among applications”.

This has several important consequences:

• ‘Best effort’ broadband has no ‘intentional semantics’, i.e. anything can happen and you’ve no grounds for complaint that the casino is at fault.
• Attempting to track back the emergent outcome to a specific mechanism’s configuration puts you into a mathematical maze from which there is no escape.
• You cannot attribute intent to any outcome, so you cannot distinguish ‘discrimination’ from ‘bad luck’.
• The ubiquitous assumption that local ‘equality’ to packets (whatever that means) results in global ‘fairness’ to users is completely false.
• Regulators can never implement ‘neutrality’. There is no objective metric, and the false positives and negatives overwhelm any attempt to establish causality.

I could write a thousand more words on packet arrival patterns and probability theory. Let’s just take it as read: a lot of people have been spouting about broadband policy without really having grasped even the most basic aspects of the system they are dealing with. This is a giant & collective intellectual faux pas.

The idea of ‘neutrality’ has looked awfully sick for a long time, and now it’s dead. Its theoretical and academic basis has perished, so it cannot support further life.

There are real problems to address

Throwing rocks at vacuous ideas derived from a mad cosmology is fun, I’ll admit. But it’s a bit of a teenage pastime, and the thrill quickly passes. As a responsible telecoms professional, I have to offer a better path that deals with the issues at hand.

Users lack the transparency they need to make informed selections of broadband products. Information about local traffic management is unhelpful for understanding the overall fitness for purpose of the service. We need something better.

ISPs do have unreasonable power to ‘pick winners’ and extort developers. Networks are resource trading spaces, and they can be both the ‘casino’ as well as ‘gambler’ at the same time. That’s not good, as it invites bad behaviour.

Developers have become accustomed to writing applications based on blind faith in the benevolence of the dice at the network performance casino. This isn’t smart, and we need a way to ease them towards reality. More of them need to signal quality needs, and pay accordingly via ‘fast lanes’ (and get low cost ‘slow lanes’ too).

A new and different debate

To address these issues, we need to start a fresh ‘post-neutrality’ debate.

The first things is to stop noisy stakeholders (often funded by content companies benefitting from the status quo) from attempting to impose their totalitarian Internet visions on everyone else.

The second thing is we need to get the science sorted out as a matter of urgency. The constraints of physics and maths can be taken as absolute and non-negotiable. Everything else has to fit within those.

Finally, we must address some very basic questions:

• What is the service that a ‘broadband service provider’ delivers, anyway?
• How can we quantify and capture that service description in a user-centric manner?
• What is the resulting role of regulators in creating and maintaining an open and transparent market with many competing technical and commercial ideas?

The future of broadband is about equality… but of people getting the service they paid for, not packets getting ‘neutral’ traffic management.