How Bitcoin's Blockchain Can Deliver Value

Blockchain is already a huge phenomenon. But it has issues. Can blockchain fulfill its promise? Can it have a big impact on financial services, healthcare and perhaps other industries? You betcha! Here's how.

Blockchain is big

First it was BitCoin. Now it's blockchain, which is what people call the "underlying technology" of BitCoin. Although that's about as accurate as calling the combustion engine the underlying technology of the car, only leaving out the gas tank -- it won't work!

In any case, expectations among investors are riding high. Here's a snapshot of investor interest:

Obstacles to blockchain's success

Probably the biggest problem with blockchain is that it's just a kind of database with a peculiar set of characteristics. Just having a database doesn't make everyone involved in a problem suddenly decide to digitize everything in the same format, make the same calls to the same API's and respond to everything appropriately. So the problem is often one which blockchain simply doesn't solve.

There's also a deep obstacle. It's one that is obvious to any technical person and makes common sense, but for reasons that elude me, is rarely discussed among blockchain investors and enthusiasts.  The problem is the community validation protocol that assures the integrity of the ledger. As part of BitCoin, the problem is cleverly solved by miners, who are incented to provide this service by being paid in BitCoin. When you take BitCoin out of "blockchain" and apply blockchain to other applications, you take away the payment mechanism for miners to do their job. This fact alone has a long string of ramifications, among which is the integrity and security of the "distributed ledger."

History of other advanced technologies

The obstacles to success for blockchain are serious. I wonder what we could learn from looking at history? Perhaps there are other abstruse, advanced technologies that also overcame serious obstacles and became smashing successes. Maybe we could learn from them and see how blockchain could similarly enjoy the great success that is so widely predicted for it.

There are several historical parallels that could be applied to block chain. One that seems relevant is neural networks.

HNC Software was started in 1986 to exploit the incredible power of neural networks. "Neural," like what's in your brain, get it? It's like smart computers! The bright boys of HNC got contracts and were solving all sorts of important problems in multiple fields. Sort of. Around 1990, the CTO of a credit card company, Household International, brought them in to solve his company's fraud problem. After a while, with the data and help of Household, they got something good going.

Household let them get data from other card companies to make their models even better, and soon HNC was rocketing. They went public, became the card industry's standard solution, and eventually merged with FICO.

I was active in the card industry when they were rapidly growing, and I knew a couple of the people whose names were on the patents. Everyone accepted that HNC's solution trumped everyone else's because, after all, it was based on that amazing new technology ... neural networks!

Looking under the covers

If you took off just the top sheet, you would see that HNC had devised carefully controlled training protocols to overcome the over-training that frequently hobbles neural networks. Clever!

But seeing what was really going on was an eye-opener. Underneath all the neural net stuff was a large and growing body of ... get ready ... human-created rules! Little snippets of code, each of which would identify a certain pattern of fraud. As new fraud was discovered, humans at HNC (mostly) would add a new rule to the set of (when I last saw it) thousands of rules. The latest set of card transactions, both legit and fraud, would then be run through the training system, and the neural network would be trained to balance the output of all the rules and combine them into a single probability score.

Net-net: the rules did the fraud recognition, the neural networks weighted the rule outputs into a combined final score for a given transaction.

The message of HNC and the understanding of the market was clear, however: the HNC solution was better than anything else, it was based on near-magical neural networks, and life is good.

Applying the lesson

HNC provides an excellent model for how blockchain can succeed in the market. They'll do it the same way neural networks did. Investment and attention will go to important problems; the problems will get solved by stubborn, hard-working people who are motivated to get something that works, and don't really care about the extent to which what people call blockchains are involved. The marketing people will promote the magical elixir of blockchain, everyone will ohhh and ahhh, and no one will care what's under the covers, so long as it works.

It worked for neural networks and HNC, and genuinely frustrated the fraudsters. The same pattern can work for blockchain.