..

expensive networks are less open source

Linus Torvalds, the creator of Linux, famously said, “Given enough eyeballs, all bugs are shallow.” This quote highlights a key principle of open-source software: the more people who scrutinize a piece of code, the quicker and more effectively bugs and issues are identified and resolved. This concept underpins the open-source movement, which advocates for open access to source code and collaborative development as a path to more reliable and higher-quality software, in contrast to traditional closed-source methods.

In the blockchain world, a related concept involves the cost of running a full node. A full node typically means downloading the historical state, verifying it up to the current chain tip, and participating in real-time block reception and validation (not necessarily proposing blocks for network consensus). The cost of running a full node varies across different crypto networks, ranging from a few hundred dollars to millions.

The expense isn’t just about hardware and maintenance; it’s also about the complexity of the node software. I believe that the cost of running a full node is generally indicative of a network’s openness or, more specifically, the applicability of Linus’s rule. Low-cost nodes, which are more numerous and accessible, invite diverse participation. In contrast, networks with prohibitively expensive nodes often see participation limited to businesses, while affordable nodes attract hobbyists.

Hobbyists are crucial. They tend to be more neutral and openly critical. When the cost to run a node is low, a network is more likely to face scrutiny and attract a broader range of observers. Conversely, in networks where node operation is costly, only businesses with different incentives and motivations tend to participate.

This situation has two primary effects:

  1. Bugs in network architecture take longer to be discovered.
  2. The network may accumulate unaddressed risks over time.

The latter point relates to a lack of incentive for whistleblowing; the former, to the simple law of large numbers.

Consequently, I’m more cautious about networks with high participation costs. They require more time to prove their robustness, and they carry a higher risk of accumulating systemic issues. While high-cost networks can still function, when it comes to the robustness stemming from open-source development, I tend to trust architectures derived from networks with lower participation costs.