Don’t Decentralise Today. Invent Decentralised Tomorrow
I am an avid collector of #blockchain SWAG and in my collection, one of the best shirts is one from my friends at EOS. It says: decentralise everything.
Unfortunately, IT DEPENDS. If you prefer, you can follow this tweetstorm — or bear with this rambling blog for a minute.
The 2 by 2 of innovation
Since I am a consultant with an MBA, true to form, I must reduce the universe into a 2 by 2 matrix of truth. The first 2x2 describes the four buckets where we can plug most innovation ideas into.
- As it turns out, creating a new business model is always risky. Let aside timing issues and execution risks, there may not be a customer for your new business model to begin with.
- Using new technology, even for the sake of greater efficiency is also risky because you may fail to execute on the best ideas if that shiny new ‘blockchain’ is not yet ready or if the regulators don’t subsidise your world saving ethanol engine.
- The riskiest thing of all is creating a new business model on emerging technology – like #bitcoin. This is why almost all altcoins died without a trace and an extraordinary fraction of tokens will also die.
- If you are engaged in neither business innovation nor technology innovation, you are either asleep on the wheel, or you are an arrogant monopoly – either of which is not interesting for this discussion.
- There is a fifth choice which is to start with a new business model and invent new technology for it — like google or amazon do, but let’s face it… most businesses can neither attract, nor retain the exceptional talent required for such miraculous. If you can do that, you are in the Genius bucket and godspeed to you for that.
- Sometimes a new technology comes along in one walk of life – eg finance, and makes it possible for entirely new businesses to be conceptualised in other areas eg transportation, that were not possible before. Self driving cars and cryptocurrency based businesses would be good examples.
What exactly is decentralisation anyway?
The popular myth is that if you put any business on a blockchain, it becomes decentralised. This tendency to rely on heuristics is not new. In 1999, people were putting Walmart, dogs, clothes, plush toys, criminal activities, books and dating — all on the internet and calling each an internet business. Sadly, putting a brick and mortar biz on the internet didn’t make a dot com then and putting a centralised business model on a blockchain doesn’t make it a decentralised business now.
All this new thoughtless push for decentralisation does is come up with stupidities like — “decentralise a central bank”. Wot? If you decentralise a central bank, its no longer a central bank. Either its a central bank, in which case it is centralised and has a considerable policy influence over money supply, currency and inflation — or it is decentralised, in which case it is at the mercy of the collective action of the acrimonious, trolling mob of miners — like bitcoin is, and no longer a central bank.
Lehman, Snowden, Equifax — the social underpinnings of decentralisation
To be fair, the cry for decentralisation is a consequence of the masses losing faith in institutions and the failure of markets and regulators to inspire confidence. Fintech thoughtleaders call this social phenomenon — loss of trust.
Wall street was the first to fall. The oligopoly of bigger and more powerful investment banks became a tool for all of us to borrow well and truly beyond our means, buying more and more stuff we didn’t need and producing so much trash that we still don’t know where to park it. We could say that our borrowing and risk taking experience was “frictionless”. At some point, we consumers could no longer pay the interest on this mountain of trash debt, and the whole system came crashing down, taking our jobs, pensions and artificially inflated living standards with it. Now a pampered modern consumer like me is nothing if not self-entitled, so we all rose up in flames, occupying wall street and regulating the street out of all profitability for over a decade.
When we lost faith in the street’s ability to give us the goodies, we sought refuge in Silicon Valley — letting Zuck and Jobs do whatever they wanted to do. We gave all our data gladly to Google, Facebook, Amazon and Apple so that they could make more and more toys for us to play with. At some point, these GAFA guys became too big and too powerful and started getting too close for comfort. Then came Eddie Snowden and Julian Assange, who drew a pretty scary picture of this matrix we had become tied in and so Europe reinvented data privacy rules with GDPR.
Of course, the self-entitled consumers we are, we couldn’t be blamed for our collective greed then, and we can’t be blamed for our Laziness now. Still, the more tech savvy among us figured that what has already happened to ‘our’ money is happening to ‘our personal data’ too… as in our time and lives are not in our control anymore. Whosoever has our data, controls our time… and our lives — and so we asked for our lives back with the cry of ‘decentralise the internet’. In other words — decentralise control of data, i.e. information.
Then, to top if all off, Equifax, the ultimate centralised custodians (or squatters/aggregators) of our identity data got hacked…
Decentralisation is about control — of money and of data
The 2 by 2 of decentralisation aka who controls the data and who gets paid for it
Distributed, Decentralised, Disintermediated
Here are some examples:
- If I own all the branches of a bank, the bank is mine, even if the branches are spread all over the globe.
- If I control all the nodes in a hadoop cluster or distributed database, it’s my database. This database may be on a million box cloud — so what.
- If I then control all the data on this database and you can’t charge me for giving it to me or take your data back — then it’s become my data.
Lots of centralised businesses today are built on decentralised architectures and with the advent of ‘decentralised ledger technology’ aka DLT, we will see more and more centralised businesses being built on decentralised technology. This is because consortia will come together and realise only too late that they still need someone to administer, govern and manage the shared platform, and take the legal liabilities for nonperformance. This somebody will become a DTCC, a SWIFT, a Euroclear — and no technology can solve that. The solutions for this problem lie solely in law and governance models, and we haven’t quite evolved as a society for that yet.
So I prefer the words disintermediation — or dis-reconciliation, which for the most part is a matter of reducing intermediaries and eliminating the insane amount of reconciliation in internal and external business processes. Incumbents can also collaborate using DLT generating billions of dollars in consumer and shareholder value. For the right use case executed well, this approach of ‘decentralising or in fact re-centralising today’ can be extremely powerful and save us many billions in today’s waste and friction. But recentralising today is not that source of new trillions of value in transformative innovation that we are all looking for – and it certainly does not always result in decentralising control of data or value.
De-centralised is not binary and it’s certainly not always better
The. most common mistake is to see centralised and decentralised as binary states. In fact ‘centralised’ is a real number between zero and one. Even bitcoin has disproportionately influential miners and too much bitcoin is in a few addresses as of today.
Decentralisation comes without recourse
If much of your net worth is tied in bitcoin, then these few addresses (maybe people, maybe not) control your financial future, if not more. It’s not that different from working for Lehman and having your net worth tied into Lehman stock when you think about it. The core devs can blow it up, the miners can blow it up, hackers can blow it up — or you can lose your keys.
Things go wrong all the time in a day’s work – whether trade finance or post trade. For most day to day business, someone’s gotta be there to be taken to court when things don’t work – or you can buy bitcoin, at your own risk, with no recourse – e.g. Mount Gox.
Only this time, you ‘fully decentralised’ – so you can’t occupy bitcoin street. There is no street.
Analytics works better when centralised
Don’t build google using a blockchain.
Google does an awful lot of things incredibly well. The most profitable thing it does is advertising. Yes, we all like to pretend that we hate those creepy, pesky ads that seem to read our thoughts. We also do find it scary how creepier these ads are getting by the day — pumping ICOs on me when I write about ICOs and sending me offers for stuff that I want but I should not touch.
Yet, we don’t stop using google. We don’t stop using facebook and we can’t stop buying google or facebook shares. Yes — the self-entitled, always complaining consumer is a split personality — we complain about what we love. Such is the internet.
Google does this magic using two things — it sucks in a mindblowing amount of data on the cloud from everywhere and it uses powerful mathematics and machines to crunch through this data incredibly fast to find patterns about my world and create a picture of me in my day to day context.
My data alone is not worth as much as I think it is
Google is able to do this because it sucks in the data first, combines it with all sorts of learning from past data and then crunches these together to improve upon the patterns it has already found. I am able to use Google in a frictionless way because Google knows my context! The frictionless experience google gives me comes from context awareness. For that google needs not just my data but also the data of the world — of all the other users, and about things we ask about. In fact, if google only crunches my data without looking at other people’s data or the data about who I work for and where I live and where I shop, google will be relatively useless. If google doesn’t know about people similar to me in a context, it’s still pretty useless to me.
All I am handing google is (a) behavioural data (b) my own reference data. In isolation with other people’s data (b) is valuable and should be owned by me but (a) is not worth much to google.
What about interstellar?
Futurists argue that decentralised technology will become so good in the future that one day it will become better than the current centralised technology in terms of scalability and performance. A commonly cited example is that of a scientific project that tries to find intelligent life outside earth using my computer when I am not using it.
Sadly, thats not a business. Even worse, it’s using stuff I paid for without paying me a dime — in a sense, this project is like centralised malware (think denial of service attack bots) used for a great and perfectly honourable purpose. If I care about finding life outside the earth, it’s brilliant and pays me in scientific and moral satisfaction. If I don’t care about life outside earth — or for argument’s sake, if my religion doesn’t want it to be there — then this software is actually fulfilling someone else’s purpose using my assets.
In that sense, this science project is a lot more decentralised than google in terms of compute and storage but a lot less decentralised than google in terms of value. Google simply takes my data for free and give me something incredibly useful for free — while charging advertisers for analytics.
As Igor Kosta points out, using a blockchain, at least in theory, you can decentralise both data and value. ‘My’ identity data can remain on my node or be shipped to the cloud only temporarily. Either Google can ship extra data to me to be crunched on my machine and combine results later; or with my permission, take my data temporarily to crunch it alongside other data, paying me a few ether or Google tokens for it.
I am awaiting examples of businesses that works like this but we can’t say that one never will. In fact, as we argue later – consumer preferences or regulation can change how the internet works today. However…
Decentralised is usually slower and less scalable
Web developers know that stateful systems scale less. This is because maintaining state creates a bottleneck somewhere or the other — either in writing data, or in reading it, or both. When this state is consensus state, the bottleneck gets bigger. You can tune it but you can NEVER get to the same scalability as a stateless version of the architecture. Stateless is the starting point and adding state slows down stateless… hence stateful is always less scalable than stateless for the same functionality.
Decentralisation works beautifully if the architecture is stateless and we are all on our own. Sadly, we are not on our own in consensus computing — this is why it is called consensus computing. A consensus state is some information about the world that enough of us agree enough on — and we collectively move from consensus state to consensus state — and a blockchain is just this distributed state machine for consensus state. The stronger this consensus and the more decentralised the control over data and decisioning, the more secure the network against malicious actors and the more trustless (i.e. trustworthy) the system becomes — and the more expensive it gets to maintain consensus.
Instead, you can hire a bank e.g. TripathiBank, pay it a fees and let it have your money in a relational database or mongo in accounts backed by a deposit insurer arm of the government. You can add some APIs on top of powerful crunching engines that analyse this database, which can be distributed over a cloud for scale, high availability and speed and the APIs can give you the functionality you want through a simple mobile phone.
It turns out that at least today, for maintaining the same level of trust and providing the same level of consumer experience, this centralised bank doesn’t need to be near a hydroelectric power plant. It can be based in London, which shut down its last thermal plant sometime back and hasn’t even got a waterfall.
If you don’t trust this Bank, you buy bitcoin and you pay the miner fees for all that electricity, and then some wallet fees etc. etc. When you lose the keys to that wallet, sorry — your problem.
Yes you can’t have it all for the same price. Sometimes you simply can’t have it your way. Life in architecture is all trade offs — sorry. You pick.
Counterpoint — Better is what the customer pays for — so always start with who the customer is, or could be
As of today, a consumer like me simply wants to be lazy. A generation of internet businesses are built around this laziness. It is also known as ‘frictionless user experience’. As it turns out, friction is poorly understood by ‘next-best-action folks’ who only look at moment to moment friction rather than over the lifecycle of the consumer.
Indeed, the Equifax hack is not a frictionless experience. If my bank gets hacked and my money gets stolen, that’s a very high friction experience. If Google hands over my data to my government and I don’t like their policies, they might put me in Jail — which is an extreme friction experience. This is why cypherpunks are willing to suffer the brutal ‘friction-fulness’ of crypto wallets etc. They don’t just care about the friction at the front end (e.g. the google search text box), we also care about the friction at the back end of the transaction life cycle. The same applies for subversives, outlaws, political activists, fringe societies and increasingly — all of us who care about privacy enough to not be so lazy.
What if
What if the consumer preferences change such that consumers no longer ‘trade front end friction for control’? 10 years ago, who thought we’d routinely sit in stranger’s cars (uber) and live in stranger’s homes (airbnb)? Now we do. What if GDPR and Equifax change consumer consciousness, and decentralised alternatives offered by Bitcoin, Blockstack and Consensys and FileCoin become GOOD ENOUGH? What if social and cultural preferences evolve to a point that we no longer care about the extra second of response time and care a lot about our most valuable assets – control of identity, time, information and money?
Laws of physics are insurmountable — but consumer preferences shift over time. So don’t decentralise central banks . Don’t try to make blockchains faster than google. Instead build breathtaking new decentralised business models — in that gambler quadrant — who knows, if your timing is right, the consumer might make you look like a genius some day. And hire a good lawyer upfront — so that you don’t land in the scam bucket.
Consumers made Zuck look like a genius — just saying.