The Dead Tokens Society

Why we Need Enterprise Blockchain Engineering

I’ve attended several blockchain conferences since 2017, when I realized that we can finally debullshitize the area by thoroughly understanding the engineering foundation of blockchain economy as a precondition to any business use cases, linking it with the already established technical excellence of the communities By the Bay. The first three conferences, CESC.io, ScalingBitcoin.org, and BPase, are strong on peer-reviewed protocol design, and software engineering of the bitcoins blockchain, the longest and oldest one in existence. The two most recent ones, Tulip and Blockchain Economic Forum, were driven from the business end, by the EOS and ICO crowds, full of investment panels and “new world models”, but both sets of events omitted a key issue.

There is indeed a key issue in the whole blockchain economy that is missing from the public discussion in both protocol-centric and ICO-centric communities. It is the elephant in the room. It shines a blinding light on the landscape of World 2.0.

How do tokens die?

Most companies I talked to at BEF share the following common characteristics:

— a great intro video with a British-accented baritone voice-overs starting with the sun rising over the globe, a company logo appearing, and large letters declaring NEW WORLD (of gaming, real estate, virtual reality, finance, you name it)

— an MVP with “global clients under NDA”, developed by an Eastern European team

— launch two quarters from now

— the CEO on the panel for great shoes

— But the presale is ongoing or already completed.

So we have millions of investors who spent billions of dollars on tokens, and the tokens were never used. And most never will be used. Ever.

At Tulip, I’ve met VCs whose whole premise is that tokens for dApps are hard to get. There should be a wallet to simplify dApp tokens exchange for BTC or ETH instantly and use the dApps. After all, if you want to play a game or go on a date or book a room, all in a decentralized fashion, you can’t wait for BTC or ETH to settle. But the disturbing fact is that most dApp owners don’t know how users will get their tokens. The dApps are not online and there’s no usage. It’s all ideational, and you might say, delusional. Or, aspirational and forward-looking investment, as the crypto VCs are telling us.

I’ve asked the BEF investment and value creation panel, that self-organized with Adam Draper moderating, what are the scenarios of value destruction, and as the investors themselves and boosters of the mass investment, what is their preparedness for such outcomes? Uncomfortable silence ensued, as if someone has just rained on a great parade. “We could die”, said Adam finally, and everyone laughed. Nobody really answered my key questions — what if the tokens never launch, or what if their issuers fold — on a technical level of any kind, or on a business level bearing any meaning. The points were made that people invest in startups, and 95% of the startups fail. Adam also said he enjoys his cryptokitties.

So, should we expect 95% of all the tokens, and corresponding investment, fail?

Let’s step back and ask the basic questions.

Are tokens like gold? When do they acquire value and when do they stop carrying any value and die?

Tokens are means to transact on the blockchain. A blockchain is a distributed system first and foremost. When we talk about a distributed ledger, that is append-only, immutable, replicated, and unalterable, we describe Apache Kafka, or any event-sourced system, that preceded the blockchains. The consensus mechanisms such as Paxos, RAFT, and other forms of BFT were already used in such systems. The key insight that launched bitcoin, proof of work, added a hurdle that validates the blocks added on the ledger, to the system, in a truly decentralized manner, opening it to the world-scale. But the ledger is not a dead record. It must live on an operational, living system. A token is a means of continuing the transactions, and the transactions must be handled by enough nodes — servers, computers, iPhones, living, beating silicon hearts with electric currents pulsing through them! — so that the promise of distributed consensus and trust network can continue.

There are several kinds of tokens. Some have their own blockchain. If they are planning to launch, chances are, they never will. Their founders, having collected the funds in an ICO, go silent, and nobody can tell us how much of the funds are already gone. Some tokens are implemented on an existing chain, such as Ethereum. But you need them flowing to mean anything.

Once we agree that a blockchain is first and foremost an operational distributed system, we have to consider the questions of how such a systems is actually run. By the Bay, we have a few companies who run distributed systems for themselves and others, including, but not limited to,

• Google — the creators of Map/Reduce, cloud, and much more
• Amazon — who took it and provided AWS as a public version to startups,
• Cloudera, HortonWorks, MapR, and other packagers of distributed OSS,
• HashiCorp, Docker, Mesosphere, a plethora of Kubernetes, enabling orchestration of such systems
• Twitter, that built Apache Mesos in-house and gave it to the world along with the Mesosphere cofounders,
• Red Hat and other Linux distributors that made Google possible and now have to deal with the consequences on all levels, providing tooling, cluster management, and even serverless plartforms
• Uber, that operates a network of millions of riders and drivers banging on their iPhones, that connect to the cloud to match them and move them where they need to go

Blockchain companies are talking about replacing VISA and becoming Economy 2.0. They want a throughput of hundreds of thousands of transactions per second. Then we can say that the performance of such networks must be at least similar to Uber and Twitter. They must use their own infrastructure or rely on the cloud providers to scale. In both cases they need the people to run the network, now called a strange word, “devops”. They are developer-operators, who manage infrastructure as code, and alas not so much as code, to make sure the “sh%t is up at all times.” They use tools like PagerDuty, that started at Amazon, to have DevOps engineers “on call” at all times to bring the system back up if it went down.

How many of the blockchain people you saw at Blockchain Expos, Forums, or Conventions are familiar with running a traditional distributed system at “
web-scale”? Try to ask some. While you are at it, ask, what programming language have they chosen to implement their blockchain? if they say Python, JavaScript, or Solidity, ask them, how are they different from Haskell, Scala or OCaml, and how do they formally prove their smart contracts are correct? And whether they know before a deployment whether the network will keep going?

When a distributed system is built to serve a business function, be it matching riders with drivers, sharing tweets with followers, or transacting in tokens, it may succumb to the following unfortunate events:

— it might never get launched in the first place, because the team was incompetent, ran out of the runway, lost motivation, or any combination of those plus more like that

— it might launch and then fail, and try to get up and fall again, and it might perform much worse than promised, to the point of unusability

— it might cease operating and die.

What will happen to the tokens issued by the ICO boosters above, whose networks failed due to the problems outlined? Will they become antique coins, appreciating in value? What will happen to the dApps that are running on such tokens?

What will happen to the users who bought the tokens to use the dApps, if there will be any?

And finally, what will happen to the investors who bought these tokens to 10x them?

If you want to think aloud with us, join RethinkTrust.org in Amsterdam, June 29.