That thing about “Y no real blockchain use cases yet tho?” popping up again. A blockchain is an accounting database that sits between accounting entities. Let me walk through this quickly.
A firm is an accounting entity. I call them accounting entities because I want to stress that firms are accountable for things: loans, equity investments, work done but not yet paid for, etc. Accounting does the accounting for such things. The thing between firms is called the market.
An accounting database records accounting data, d’oh! Accounting data are the entries about business transactions and transfers that sit between the finance and operations layers. A sack of rice (OP) for a bag of coins (FIN).
[NB database transaction ≠ accounting transaction.]
Two things about accounting databases:
1. they come with an inbuilt ad-hoc validation process, namely debit = credit, and 2. they record all business transactions in a single unit, the unit of account, so the sack of rice is recorded in the unit of the bag of coins.
For the last, I don’t know, couple thousand years we have kept accounting records where they were needed: within the accounting entity, from clay tablets to big irons. So why even bother flipping this over now? Mind blown.
So are blockchains really just solutions to problems we never had? Or are they solutions to problems we never realized we had, until someone pointed them out to us?
Turns out the Satoshi whitepaper was a bit of a “Sex Pistols at the Lesser Free Trade Hall” moment for enterprise databases. It took the Pistols a couple of years to play Wembley Stadium.
Wait, they never did, but…
After a couple of years it dawned on the enterprise software makers that “blockchain” could potentially be the solution to a problem their customers have been banging on about for a while, the whole “data in silos” problem.
Of course that problem is entirely self-inflicted. If you build your enterprise data concept around the enterprise as an accounting entity, you’ll do all kinds of things to keep your enterprise data from leaking out.
Especially since the whole Satoshi/bitcoin thing was couched in rather anti-enterprise terms and kinda sketchy and illicit anyway, and also completely overengineered to solve a problem enterprises don’t really have the decider folks didn’t pay attention until fairly recently.
So the whole blockchain-not-bitcoin movement 1. isn’t that old to begin with, 2. didn’t get C-level interest (and budget) until fairly recently, 3. is still mostly in the “innovation” phase, and 4. somehow still has to fit into the whole enterprise-centric product portfolio.
The last point might be the biggest reason for the slow adoption within enterprises. If you spent the last 40-ish years building enterprise databases and processes, wrapping your head around the fact that all of this can be flipped over is a bit mindblowing.
You run into problems you never thought you’d had. You run into the simple problem that you don’t have the people who can solve the problems you just discovered. You run into the problem that the use cases that looked promising from afar don’t pan out on closer inspection.
Just to pick a very fundamental problem. Enterprise planning runs on decision theory. You want to aggregate information and set decision points to optimize for a single entity: the enterprise. “Blockchain” planning runs on game theory.
So you see most enterprise players retrenching to build “blockchains” that have very little to do with Satoshi-model blockchains, triggering the cypherpunk commentariat to decry: “Those aren’t real blockchains!” Fuck them.
Quick detour: autonomous driving research has been around since the mid 1980s. Even with big money and early successes the topic disappeared from the radar between mid 1990 and the mid 2000s, when DARPA kicked off the Grand Challenges and cars drove into a ditch after 7 miles.
Since then interest and funding have been off the charts again, but 14 years after the first DARPA Challenge there’s still no autonomous car in production. Y tho? Bc momentous inventions have momentous legal challenges, something cypherpunks love to ignore.
A result of these legal challenges was the automated (sic!) driving taxonomy that eventually got adopted by the SAE. The gist: we are moving forward in stages, and we’re only going truly autonomous if the risks can be contained. Getting from recognition to adoption took about 10–15 years.
The enterprise blockchain world is converging to the same gradual approach, even though they haven’t codified it yet. That “Ahh that’s not really a blockchain” database might indeed be a level 2 blockchain, moving some enterprise stuff to the market.
In 1998, Paul Krugman didn’t just predict that the internet is toast, but also that information technology is overrated because word processing was still the same as in 1988. Apparently he’s never seen a backend or realized how enterprise systems affect the stuff he’s working on.
To paraphrase Amber Baldet, one of the most level-headed people in the cryptoverse: the rush to productization can lead to crappy products. I mean I love “lean startup” and throwing crappy stuff on the market to see what sticks, but it has its scope.
And when humans can get hurt by your crap, or your recalcitrant refusal to think through the knotty problems that make your stuff crap, then your invention is outside of that scope. Lessons learned the hard way. ◀️
▶ I realize I wanted to say something about “Are there any use cases at all tho?” Short answer: yes. There are many cases where an in-between-entities source system of record can make sense eventually, and much fewer where it makes sense now.
Enterprise is just about to move into the cloud, starting to re-think the boundaries between transactional and analytical, between mutable and immutable, between systems of record and systems of engagement. Legacy systems and problems abound.
At the same time outsourcing continues while supply chain becomes ever more complex and “scope” and “speed” replace “scale” as value drivers. Shared source systems of record can be a godsend for that.
Cryptocurrencies might have potential for payment automation eventually, right now nobody wants to touch them bc they’re an accounting disaster.
So no, the first generation “enterprise blockchains” will not look much like Satoshi blockchains. That’s fine as long as they implement 1. shared control, and 2. ad-hoc validation of some sort.
On the other hand, cryptocurrency blockchains better not look like Satoshi blockchains either in five years. Maybe the twain shall meet again at some point after all…
The original tweetstorm appeared on @ecoinomia. The text has been slightly fixed and it might still evolve into a real essay. Oliver Beige is an industrial engineer turned economist (PhD Berkeley, MBA Illinois, MSIE Karlsruhe) who is focusing on how technologies like machine learning or blockchain might change the value creation between markets and enterprises.