Johnston’s Law Quantified

“Every 2 years the number of actions relying on decentralized consensus will double.” David A. Johnston

Introduction:

In 2014 I offered Johnston’s Law as: “Everything that can be decentralized, will be decentralized”. This proved a popular and very quotable idea in the bitcoin and broader decentralization movement.

The questions I often got back were: Which use cases belong in the “Everything that can be decentralized” category? How does one quantify this increasing decentralization? How will we measure if Johnston’s Law is true over time?

After all, there are many decentralized systems from BitTorrent for sharing files to Ethereum for executing smart contracts and many metrics within each of these systems to choose from. In an effort to answer these questions and quantify increasing decentralization I selected the most generic and low level metric available “an action”. For the Bitcoin Blockchain that would be making an entry / transaction in the distributed ledger, for BitTorrent or the Safe Network that would be writing or reading a file on the Distributed Hash Table and so forth.

As a starting place for this quantification I picked the Bitcoin Blockchain to pull statistical data from because of the 6 years of data available at a day by day granularity. In addition Bitcoin’s user friendly / transparent block explorers made the process straight forward for gathering statistics.

Below I’ve completed a statistical analysis of the number of entries recorded in the bitcoin blockchain over the 78 months of its existence, from January 3rd of 2009 to June 22nd 2015. My core conclusion, based on this statistical analysis, is that the number of entries per day conducted on the bitcoin blockchain is likely to at least double every two years into the foreseeable future and this is a good rule of thumb for other distributed consensus systems.

I believe this is a good model for quantifying the increasing use cases that rely on the bitcoin blockchain and the general utility of decentralized systems beyond that of the bitcoin blockchain.

Here is the link to the raw data and statistical analysis I conducted, available in a publicly viewable Google Sheet.

https://docs.google.com/spreadsheets/d/1OtvLuUJmy4seXt2Rc8fskbTuGRIOtbAgD_QwdCE2Fbc/edit?usp=sharing

Analysis of Trends In The Data:

During the first year of the blockchain’s existence (2009), it basically existed as a proof of concept and experimental system, which was primarily used by cryptography researchers and a small group of dedicated technically savvy early adopters.

As you can see from the screen shot below, the bitcoin blockchain started to experience rapid growth in usage from early 2010 and continued on this trajectory through mid 2013. During this period the number of entries grew 10% to 60% per month. This period was marked by increased global awareness about the invention of the bitcoin blockchain and broad media coverage of the topic, especially the use of bitcoin as payment in online commerce.

Toward the end of 2013, growth in entries fell to less than 1% per month. Growth levels did not recover to above 5% per month until the first half of 2015. This period clearly corresponds with the collapse of Mt. Gox, which in late 2013 was responsible for a majority of the world’s bitcoin exchanges for fiat currency and thus its users were responsible for a sizable number of the entries being made in the Bitcoin Blockchain. Since the Mt. Gox collapse a number of new / competing exchanges have risen to fill the void, including Coinbase, ItBit, BitStamp, BitFinex, BTC China and many others who have more technically robust, regularly audited and financially scalable systems capable of supporting far greater exchange volumes. In addition use cases for the Bitcoin Blockchain have expanded to include meta data entries representing ownership in tokens, gold, stocks, bonds, titles, and more.

Stages of the Bitcoin Blockchain Growth:

Broadly speaking thus far bitcoin blockchain technology has experienced three distinct stages of growth.

1. Infancy during the experimental days of 2009.
2. Hyper growth while the world first learned about the bitcoin blockchain.
3. Consolidation and a focus on scalability in technology.

Future stages of growth in bitcoin blockchain usage will likely be heavily impacted as developments unfold in the next 12 months. For example if the block size limit remains at 1 MB, then as transaction growth hits the limit (likely during the first half of 2016) use cases unable to pay for higher entry costs (or if the limit causes confirmation delays beyond the typical 10 minute period), will either move to competing blockchains or “off blockchain” all together.

On the technical side, the introduction of Side Chains for coin transfers and Factom for data publishing connected to the bitcoin blockchain, will have a hard to predict impact on the number of entries recorded on the bitcoin blockchain or via these new layered technical solutions. Should these two layers become popular, then in the future I would include statistics from these layers as entries being made on the bitcoin blockchain, as they are clearly using the bitcoin blockchain as the root of their security model.

The Idea of Scarce Blockchain and Distributed Consensus “Capacity”

Instead of asking “How many actions / entries will users make using distributed ledgers?” perhaps another approach would be to ask the question, “How many actions / entries do distributed ledgers have the capacity to support?” There has been extensive discussion (as part of the maximum block size debate) around the fundamental questions of how large the blockchain can grow, while still remaining “decentralized”. In this context developers are debating how many actions / entries the average bitcoin node with an average bandwidth connection has the capacity to support.

From this angle we can think of distributed ledgers as having a very real limit to their capacity to record new actions. BitTorrent nodes for example are throttled based on the resources they add to the network, similarly the Bitcoin Blockchain clients place a cap on the size of each block of transactions (currently 1 mega byte). The doubling in this context is similar to the doubling of computational power in that users will likely always find more uses for the available distributed ledger capacity. Thus usage of the blockchain over time is only constrained by the scarcity of that distributed ledger.

If proposals for Bitcoin’s core client adopt a biannual doubling time for the block size cap (see Gavin’s Bitcoin Improvement Proposal 101) this may become a self fulfilling prophesy where the broader industry sets its growth plans around the expected future capacity of the bitcoin blockchain. This would be a similar phenomenon to what has been seen in the computer chip industry. Some people speculate CPU development the past 50 years has followed “Moore’s Law” in part because the industry itself uses the predicted 2 year doubling time in computational power, as a benchmark for their timelines around designing and fabricating their new computer chips.

Other Metrics:

There are many metrics and ways of measuring decentralization and the bitcoin blockchain in particular that could be selected for this type of analysis, including total value transferred, number of wallets / users, number of addresses and so forth.

However, at its core the bitcoin blockchain as a distributed ledger is keeping track of a series of entries that take place. Given there is little realistic possibility of quantifying the amounts of value recorded / transferred using bitcoin blockchain entries (representing values as diverse as currencies, stocks, bonds, real estate, digital tokens + innumerable other use cases), the number of general entries on the bitcoin blockchain ledger seems the best available measure of its generalized usage and utility.

Quantifying Other Blockchains & Distributed Consensus Models

So far I’ve only discussed the bitcoin blockchain. But there are many blockchains under development such as Ethereum, and Distributed Hash Tables are used by projects such as BitTorrent and the SAFE Network.

Clearly each of these qualify as a “decentralized” adding to the trend toward decentralized systems and applications, but the question remains on how to quantify their increasing adoption. Should smart contracts on Ethereum or files on BitTorrent or the SAFE Network count toward the number of entries in distributed systems? To me it seems clear that they should be, however today in the case of BitTorrent for example, system wide statistics are difficult to determine and verify.

I think it will become possible over time as technologies mature and layers such as Side Chains and Factom increasingly weave together all the competing blockchains and distributed systems (making them more interoperable), then these systems will also end up themselves making entries into the bitcoin blockchain. When these systems become interoperable, then their actions and entries will become easier to quantify and should be counted toward the trend in a more unified way.

Jeff Garzik stated it well when he said that the bitcoin blockchain increasingly serves as the “root of security” for other blockchains and related decentralized systems.

Relationship To Moore’s Law

Now that the world’s companies and individuals know about these distributed consensus technologies, they are increasingly experimenting with it for their own use cases including huge institutions such as UBS, USAA, Citigroup, NASDAQ, IBM, ING, ABN Amro, and Rabobank.

Given this, it seems only logical that the usage of the bitcoin blockchain and distributed systems more broadly, should closely follow other technological growth trends in computational power, bandwidth, and data storage, which also double roughly every two years, see Moore’s Law, Nielsen’s Law and Kryder’s Law respectively.

While periods of hyper growth during the early popularization phase of these distributed systems seems common, after maturity and saturation of the early market, it’s logical to conclude that these distributed systems can not out pace the growth rate of the physical hardware (computational power, data storage and bandwidth) on which their functions depend.

Conclusions:

Blockchain technology is one of the most widely used decentralized platforms today and has proven to be anti fragile and has grown stronger with each challenge to its growth. The past 78 months it has weathered innumerable technical attacks, bad actors using it for illegal activity, unprofessional early intermediaries, a world full of critics and well funded naysayers (who have declared the bitcoin blockchain dead 71 times since 2009).

Despite all these challenges, the decentralized bitcoin blockchain and distributed consensus models more broadly offer novel and compelling new capabilities that the world of computer science has never seen before. The existence of a world wide ledger, which records consensus in a decentralized way and is censorship resistant, is certainly here to stay.

So in my quest to quantify Johnston’s Law (“Everything that can be decentralized, will be decentralized”), I think it can be best expressed in saying that: “Every 2 years the number of actions relying on decentralized consensus will double.” Thus decentralized systems will increasingly expand the usage and variety of use cases that can become trustless, peer to peer and otherwise replace existing intermediaries, thanks to their unique capabilities.

I believe this is the best measure for if Johnston’s Law proves correct over time. Conversely a decline in the use of decentralized systems relying on distributed consensus mechanisms (or future iterations or related technologies) would suggest that decentralization is not a inevitable and Johnston’s Law isn’t correct.

Only time will tell if Johnston’s Law proves true or false.