What decentralization is good for (part 2): Platform Risk

Continuing on the theme of what decentralization is good for, this week I would like to focus on one of the most powerful drivers in the near-term: Platform Risk.

Platform Risk is is the risk that the tech platform that you build your product/app/business/life on will become a critical dependency, will become unreliable, and/or worse, will screw you in the end. 

Here is a post from a few years back that details many different flavors of platform risk, many of which are benign, and some of which are malicious. And here are some examples, to make it more concrete:

This is not to say that any of these acts are necessarily illegal, or even immoral.  But if you are investing serious time and money — especially dropping everything to build a business on a platform — these kinds of risks are of grave concern.

So, what does decentralization have to do with platform risk?  When the platform is a protocol (i.e, decentralized) rather than a company (i.e., centralized), the rules of engagement are known up front and can’t change on a whim or because of a business decision.

If we think about the original internet protocols (TCP/IP, HTTP, SMTP, FTP, SSL, etc), they are a set of networking, communications and data exchange protocols that ultimate form the platform we know of as the web. While there are certain forms of platform risk on the web (e.g., stability, speed, security), the web on the whole has become a very stable and reliable platform, generally absent of the flavors of risks detailed above.  

Cryptonetworks (i.e., public blockchains and cryptocurrencies) combine the architecture of the original internet protocols with the functionality of today’s corporate applications platforms (data management & transactions).  While there are still major issues to solve before these systems collectively become a mainstream platform, they are gaining major adoption from developers in large part because developers are so keenly aware of platform risk, and see cryptonetworks as a type of platform they can trust.  

As an illustrative example, let’s compare downloads of the Truffle framework (a popular dev tool for Ethereum):

source: Truffle Dashboard

… with the price of ETH over the same time period:

source: Messari

Developers are the canary in the coal mine when it comes to platforms.  And at the moment, they are pointing to the desire for platforms with less inherent risk, more reliability and more trust.

20 comments on “What decentralization is good for (part 2): Platform Risk”

Are the Truffle download stats verified (audited)? So much shit goes on in crypto that i have ask.

the psychology of assumption. the tendency to naively adopt the mindset that because crypto is promoted as the tech of trust and truth we unquestioningly accept that what we are seeing is authentic. There are clever (but cynical) people out there exploiting that tendency. Have you ever been to Bushwick? ;-)

These are open download stats provided by npm, the preferred distribution channel for all javascript projects. You can find a version of the graph above updated daily here: https://truffleframework.com/dashboard. If you check the network activity through the Chrome Developer tool, you can see that your computer is pulling these stats directly from npm without going through our servers.

As far as authenticity, it’s definitely true that 1 download != 1 user — a single person might download it multiple times. They’re also increased by CI/CD usage of Truffle and npm mirrors ensuring they have a copy of the package to distribute to clients.

It’s likely that we can chip away some of the downloads by analyzing authenticity, which would change the number a bit, but the trend would still be the same. We also have other stats (usage of Ganache UI, website statistics, etc.) that corroborate the same usage and trend.

Agree with you about crypto being able to change auditability here. It’ll be a long road before crypto takes over all of infrastructure, but we’ll get there. :)

Thanks. Crypto will eat the world, eventually. I like the Truffles name. I’m not so sure about the logo. I like ‘associative’ identity. Truffles = wild boar (in my mind at least). http://www.hampsten.com ‘Truffle Hunters’ – a name for one of your events, maybe.

Platform risk from Microsoft and Windows?

Hmm ….

We remember when Ballmer was screaming “Developers!!!”. Microsoft needs developers of applications on Windows, either server side with Windows Server or client side with Windows, say, 10 Home Edition.

So, has Microsoft made life hard for, say, Mozilla and Firefox? For video player VLC? For some of the companies that are big users of Windows Server?

IIRC recently Microsoft has been getting about $5 billion a month in revenue. I have to believe that only a little of that is from their main client side operating system Windows 10 Home Edition. Instead I have to suspect that they are getting a lot from licenses for Windows Server, SQL Server, etc. and site licenses for Office, etc.

A company paying big bucks each month for Windows Server or 10 will expect good support for all the documentation and APIs and more support for specific problems.

So, if want to be distributed with server farms in Connecticut, Virginia, San Jose, Ireland, England, Paris, and Berlin, with all of them based on Microsoft’s products, then definitely will want Microsoft to be a fully reliable vendor with no withholding, competing with their own customers, monkey business, etc. I’m betting they will be a fully reliable vendor. Else people will be converting to Linux maybe using .NET Core there.

Net, being distributed by having several server farms with, say, some load leveling among them can help reliability but not protect against the supposed monkey business from one vendor used on all the server farms.

The time I visited a colocation site in Wappingers Falls, they explained that they routinely supply dual, quite independent connections to the Internet backbone. When I visited the NASDAQ farm in Trumbull, CT, they explained some of the extreme parallel and redundancy approaches they were taking, e.g., multiple sites, at each site, multiple, independent electric power suppliers, battery back up power, multiple Diesel generators for charging the batteries, Non-Stop computers, etc. And no big need for anything like block chain, distributed or not.

If my startup is successful, then it will need quite a lot from SQL Server. My first looks showed that SQL Server can be parallel, redundant, and distributed, all without anything like block chain.

From a fast Google search, at

https://azure.microsoft.com/en-us/blog/three-reasons-why-windows-server-and-sql-server-customers-continue-to-choose-azure/

I found

For the past 25 years, companies of every size have trusted Windows Server and SQL Server to run their business-critical workloads. In fact, more than 70 percent of on-premises server workloads today run on Windows Server1.

And I saw more evidence that serious business server farms commonly depend on Windows Server and SQL Server.

Or, if Microsoft is not a reliable vendor, then no number of distributed server farms with blockchain will solve the problem of a bad vendor for all the farms.

microsoft is a big company… so no doubt some or many of their products are for sure in the reliable / sustainable / trustworthy category

I think to a degree there is a difference between an infrastructure product like SQL and an application platform like windows or iOS, where a lot more misalignments can occur

Nick! We use to work together way back when (Open Planning Project). I created Truffle. Love that you included our stats here. Feel free to reach out if you have questions.

Great points. These issues are not well enough understood. But your post highlights that platform risk exists even where the underlying architecture is designed in a way to mitigate it. The stability and openness of Web protocols did not prevent Amazon and Google acquiring great platform power – indeed they had the opposite effect. New intermediaries were needed to manage the surfeit of choice enabled by the open Web, which benefited from network effects, and we know how that ends. Will applications built on blockchain not in the end be subject to the same centripetal forces, and if not why not?

points of centralisation in decentralised networks. that’s the scramble in the web tech VC world, to identify where those points will form. It’s hard to make outsize returns in ecosystems that promote a diffuse distribution of value. that’s capitalism’s nemesis right there.

The commercial incentive to find ways to centralise has proved pretty unstoppable (at least for the last 10,000 years of human economic life)

and modern capitalism is only 500 years old, and the VC industry +60 years. There’s definitely a dynamic force for aggregation (we’re both living on the surface of a product of that force), but if we subtract financial capital as a force for centralisation i wonder what forces remain that explain the dynamic?

Comments are closed.