The newness of broadband means our ideas of how packet networks should be built have yet to withstand the test of time. How can we know if our professional knowledge is wisdom or folly?
Whilst I was at the Welsh Slate Museum I heard a tale that made me think of the broadband business. It was the end of the day on a quiet week, and we nearly had the place to ourselves. A genial gentleman did a demonstration of splitting slate.
This was once a very dangerous job, due to the copious amounts of lethal dust it produced. Apprentices would start at the age of 12, and they would then endure 7 years of training just to do the “dressing” of the finished slate. This process trimmed the edges into the right size, shape and finish. Only once this was mastered would were they allowed to begin the splitting process.
This particular craftsman was the product of six successive generations of slate quarrymen. He certainly knew his material by the way he handled it, which reminded me of how a professional musician holds their instrument. His thickened fingers delicately pointed out how the slate has a natural grain. It is very strong when stressed along this grain, and weak when bent across it. That means roof slates are directional. You need to have them the right way around and finished the right way so that they don’t wick the water up and around the slate.
He told a tale of driving in the north of Wales, and saw a house built with a “do it yourself” roof. He noticed that they had not only failed to dress the slate properly, as it still had perfectly square edges, but also the slates had the grain going the wrong way.
He stopped and talked to the owner. “You have your slates upside-down, mate!” – “No we don’t. Anyhow, they look better this way!” So he shrugged and moved on. Why argue with someone who doesn’t know that they don’t know what they are doing? A few weeks later he drove past again. Clearly this person had decided to get some advice, as the whole roof was slowly being removed to replace the slates the right way up.
Broadband doesn’t have sixth generation network architects. It is a new industry, and we are mostly first-generation craftspeople hewing our bandwidth from glass. Some even dare roof their networks with copper, apparently! I don’t know how that stands up to the weather.
Given the youth of our industry, how do we actually know what we know? And how can you tell if your network has the “slates” the wrong way round?
I attended a conference a few years back and in the closing raffle I won a book “On being certain”. It was an entertaining read, and the message is quite simple: “knowing” is just a feeling we have, rather like excitement or fear. That means the “feeling of knowing” is not the same as something being true! Indeed, it is thoroughly unreliable. “Actual knowing” is a result of the scientific method (i.e. applied philosophy) and a lack of attachment to the “feeling of knowing”.
In other words, “actual knowing” is the exact opposite of the intuitive gut feel we get. This is why I am not an industry analyst, but instead am a computer scientist. I need to ploddingly pull apart a problem into its elemental parts rather that confer with my intestines. In fact, I’m a pretty rotten analyst, which is why I always work with other people who are good at it when necessary.
The newness of broadband means that some ideas that are widely and dearly held are wrong. For instance, consider the belief that more quantity automatically equals better quality. This is often expressed with a sigh as a wish for everyone to have a hundred gigabits to the home (symmetric, please). And on the face of it this would appear to be wonderful.
Yet it doesn’t take account of how those flows interact, or the scheduling limits of the overall stochastic system. Indeed, it is common to be unaware that broadband involves statistical multiplexing, or what this implies.
As a thought experiment, imagine a network where everyone has a gigabit to the home that goes to one of two Mothers Of All Switches in the country, which are connected with a terabit link. It would only take a thousand users to saturate the national backbone, and to perform a denial-of-service attack on the whole infrastructure. So maybe arming every user with a broadband bazooka to hose their neighbours isn’t the best idea!
The sad truth is that it is common (if not normal) for broadband networks to be “upside-down”, and to have the wrong “dressing”. As a result they are low quality structures which badly leak “user experience” and need far more servicing and upgrading than they ought to.
For the latest fresh thinking on telecommunications, please sign up for the free Geddes newsletter.