Are you a schmuck ripe for ripping off at the telecoms network casino? If you want to win at a game of chance, it’s really helpful to know which side you are on, the rules, and the odds. The abandonment of proper engineering has left the telecoms industry wide open to attack by people playing “network day trader”, armed with a better understanding of the rules, and who tip the odds in their own favour. Like me, for instance.
“Statistical thinking will one day be as necessary for efficient citizenship as the ability to read and write.” — H.G.Wells
I used to have a really frigging stupid idea. It took me seven years of failing at it to realise how absolutely dumb it was. It pains me to even say this to you.
I tried to sell advanced network quality measurement and packet scheduling technology to telcos.
Of course, it didn’t work. I know this is not an affliction that is unique to me. Many of you are busy selling boxes to operators. Some of you even make the sale, and get paid a nice living for it. But this is different.
By “advanced technology”, I mean stuff that the well-funded R&D labs of the top equipment vendors couldn’t come close to replicating. This technology falls into the “indistinguishable from magic” class. It’s beyond what the consensus reality sees as possible in terms of visibility and control over the delivered experience.
So I’ve now changed strategy, at long last. (You get to call yourself a “strategy expert” if you notice your strategy isn’t working in the first 7 years and switch.) I finally know what my job is! Would have been nice if I’d figured it out right after I took a 90% pay cut after leaving the corporate world. But there you go, life doesn’t always make things easy.
Now, pay attention, as I’m going to reveal to you the big secret of how to make money at telecoms. By this point it won’t surprise you that the essential insight is to never, ever, try to sell advanced technology to telcos. That’s totally cretinous and imbecilic.
What you need to do instead is to get a dirty great big “digital vacuum” and hoover up all their profits, and do it from the edge of the network. It’s a kind of industrialised version of going “over the top”. Just bring an intellectual nuclear weapon to this gunfight, to be sure of winning. Unfair, but fun.
What you really need to know about telecoms is that it went through a big change in the last few decades. Once upon a time, it was an industry built upon rigorous scientific and engineering foundations. People who actually paid attention during their undergraduate degrees, and even had the perseverance to do PhDs, went on to places like Bell Labs or national R&D institutions.
These engineers had a ponderous and slow process of forming standards underpinned by strong theory and mathematics. They took responsibility for their work, and the consequences of error or failure. It was a professional ethos for a service that formed critical national infrastructure.
Then in the 1970s we discovered packet networking, and started experimenting with a new digital drug: statistical multiplexing. What this did was turn networks from being boringly predictable (and expensive) services into something unpredictable (and inexpensive). But boy, could you “pile them high and sell it cheap” and re-share the underlying telecoms assets over and over.
This digital drug was so powerful it could make you instantly rich. And in our society, rich people are good and worthy people, and they certainly must know what they are doing, and it’s never ever an accident of timing and technology. So the absence of rigorous science and performance engineering was all a bit of a “meh!”.
In its place was a glorious doctrine of “best effort”, in which mysterious and magical things happened in the network (just never ask for the safety case). People find it easy to believe in things they want to believe in, and so this digital dogma caught on really quickly. It was especially good as a way to goad those silly old fools who insisted on proper engineering.
There was only one itsy bitsy catch. The statistical multiplexing game turned every packet network into one big casino where you could bet on getting good quality for each packet you sent. Even better, a horrendous (yet universal) pricing blunder let you place unlimited bets for free! Oh, and telcos didn’t properly understand the rules of the game, or the odds.
And thus came the first wave of “over the top” players, who arbitraged connectivity and performance. That cycle is now coming to and end. See my endless writings stretching back fifteen years on the matter. As I’ve switched sides, I’m now in the business of attacking the telecoms industry vigorously, and transferring its wealth into the cloud and computing ecosystem. Nice work if you can get it.
How to do this, you ask? Well, the “packet poker” game is still evolving, and to keep winning at the network casino with “OTT 2.0”, you need to improve your strategy. This means manipulating the odds by understanding the hidden weakness of your network operator opponent.
Telcos have two basic failures to understand their own business, and anyone can exploit these. Now, this is an ethical writer with a strong sense of morality. So if you follow my advice, I only have one ask. Please light a candle in memory of the pension funds you are transferring into your own pocket via foolhardy investments in telecoms capex.
The first failing of telcos is that they are sluggish. I don’t mean that they create an awful squishy feeling between your toes as you walk barefoot in the garden. It is that they work in a business where the rate at which they sample the customer experience is orders of magnitude slow that the processes that form the experience. So they’re basically blind to what’s going on at short timescales.
The second failing of telcos is that they are amnesiacs. You might thing grandma has a bad memory, but this is of a whole different order. All kinds of stuff flows through their networks, and they forget what just passed through immediately. It doesn’t leave the slightest impression on their memory, and thus learning is impossible.
So to win at telecoms (without being a telco that’s part of a spectrum cartel of badly-regulated former monopoly incumbent or franchisee) is to move faster than the telco can, and remember what happened.
You can think of this as two separate, but closely related, ideas.
The first key idea is that networks are not like ‘pipes’, but instead are more like a financial services ‘trading platform’, doing ‘resource swaps’. These systems are vulnerable to people like me, who come along with high-frequency trading algorithms that can outwit the quality rationing QoS algorithms of the telco. In terms of the network casino, it is having “blink or you’ll miss it” reactions in a game where the other participants are all stoned.
The second idea is that you can “count cards”. Now, a human card counter faces the problem of memorising the state of a maximum of 52 cards in a deck. In a telecoms network we’re dealing with millions and trillions of packets, so we can’t remember when and where everything went. Instead, we need to capture the right essence of “ghosts of packets past”, in a compact and useful form.
The bottom line is that both of these are applied mathematics problems and are totally solved. You too can outwit any packet network and arbitrage the assets, sucking out the profit! Arm the fuse on that network nuke, Bob! Oh, and your equipment vendors don’t have the maths, and don’t yet know how to engage in counter-measures (and they do exist).
So maybe a few years from now I’ll be selling anti-arbitrage systems to telcos. And they may even buy some. But I’ll only do it as a hobby, once I’ve transferred a few pension funds into my own (and my start-up investors’) pockets. Anyone got a box of candles?
For the latest fresh thinking on telecommunications, please sign up for the free Geddes newsletter.
Read It Now
