Time Be Time

There is a character in William Gibson's Neuromancer (read it!) who responds with the words "Time be time" when asked if he has a watch. I've always liked that. I once had a client, many years ago, who asked me what their data looked like until the end of the month. It was surprisingly difficult to explain that because the month hadn't yet finished, the data didn't exist.

Data is a simple concept. Events happen, somebody records things about them, and that's data. It is always past-tense. It might be data about five years ago, or five nanoseconds, but it always comes from the past. People like me use it to form uncertain assessments about the present and the future, and we can usually get a good sense of how uncertain we are. It annoys me that so many of my peers are dishonest about that.

Let me warp your mind. Data is the inverse to credit. While data always comes from the past, you can only ever borrow from the future. In the simplest sense, that's your own future. You say to a bank "I don't have enough money right now, but I will later," the bank lends to you, and your future self pays it back. You can also borrow against your estate, which effectively means your children pay it back. And, on a national scale, governments borrow from future generations of their people. It doesn't matter who the actual lending institutions are, it is future generations who have to suffer the economic harm of servicing the debt. Time be time.

There is something perverse about using data to justify and rationalize such long-term borrowing. I say that with my tongue in my cheek, my foot in my mouth, and my head up my arse. I am a data scientist, I understand that if borrowing is going to happen (and it will), then data is the only thing that can justify and rationalize it. But using the past to predict generations into the future is nonsense. We're not that good.

I like to tell people that, as a data scientist, I can predict that the height of a nearby tree will be three feet by tomorrow. When they ask how such obvious rubbish can be true, I tell them I'll attack the tree with a chainsaw. You can always predict the future if you control it. If somebody else controls it, you're dealing with uncertainty. And if the future will be constructed from millions of people's decisions over the next century — like a nation's economic future — you're stuffed. Forget it.

If every data scientist on Earth put our heads together, we couldn't tell you if a country like Spain will exist as a single geopolitical entity in 100 years. We couldn't tell you if Germany will be rich or poor. Or if India will become a superpower, be devastated in war, experience a climate-change collapse of the monsoon rains, or stay much as it is now. And we sure as hell can't tell you what happens if a nuclear power decides to default on its debts.

You know that saying, 'If you owe a $1000, you have a problem. But if you owe $1 million, your creditors have a problem'? The five recognized nuclear powers, each sharing veto power in the UN, owe $55 trillion. $48 trillion of that is owed by the old superpower and the new — the United States and China. Their creditors have a massive problem, and since any economic harm to those megaliths will impoverish the globe, their creditors are all of us.

Right now, there are data scientists advising all sorts of banks, pension funds, and the like, to lend money to those governments. This is based on their predictions about very long-term economic prospects. Those predictions are crap. Time be time.