Category Archives: Safe to Fail Probes

Safe to Fail Probes

The freedom to fail

In video games learning takes place because the players are not afraid to die. As a result of this freedom to fail a lot of experimentation takes place and over time learning takes place. Hackers tend to think in a similar fashion and the art of “fuzzing” is similar in that it attempts to make the system fail by trying to push the boundaries of the application design. In essence these are both “brute force” solution discovery methods where essentially trial and error are used in a systematic way. It may take many failures to achieve a success which is why there has to be a freedom to fail.

What are safe to fail probes?

The safe to fail probe is an experimental technique which can be applied in complex adaptive systems to discover new knowledge. Dave Snowden pioneered safe to fail probes and one of the key points is that each experiment should be different. It’s important to not have 20 out of 100 people all trying the same experiment.

Blockchain based distributed autonomous communities are not inherently chaotic

Despite how things may seem in the media the blockchain based decentralized community is not inherently chaotic. It is a complex adaptive system where order exists but it is too complicated to deal with using the top down approach. It is inflexible to think of blockchain communities as corporations because typically a corporation is a top down hierarchy, has public leadership roles, and is fairly predictable in how it operates with changes taking place rather slowly. The decentralized communities do not operate like corporations but at the same time are not “anarchy” in the sense that it’s a lawless “wild west”.  The law of the decentralized community is encoded into the software that each individual must run.

To think of each blockchain and or DApp as an experiment is to allow for the idea that a blockchain can be like a safe to fail probe.  The only way to discover what is or isn’t possible in a completely new environment is to probe the environment through multiple simultaneous experiments. So just as distributed sousveillance can allow a blockchain to probe an environment to create collective intelligence a similar approach can be used to decentralize knowledge generation through safe to fail probes.

Blockchains are distributed sensor networks

This will become clear as prediction markets utilizing distributed oracles show that the consensus (Schelling point/focual point) of truth can be identified. Methods such as using prediction markets, powered by zero knowledge proofs, oracles, connected to a blockchain are quite powerful. This means perspective is valuable for determining the truth of any event and the blockchain can combine globally distributed perspectives of an event or situation in a quantifiable way. Additionally this has implications for distributed autonomous security systems as well which at this time have not been fully explored or exploited.

A problem with blockchain maximalism (one blockchain to rule them all)

One of the major problems with blockchain maximalism is that while it does promote the network effect of the winning blockchain it doesn’t actually contribute to the process which led to the innovation of the blockchain itself. In an environment where there is a diversity of blockchains where many attempts are being made to innovate or to find solutions to previously unsolved problems then you have a way to experimentally probe the space which I’ll call the “search space” to find the optimal solution to a problem.  It allows more developers (and more minds in general) to attack the same problem with similar economic incentives to the incentives of the early Bitcoin adopters had.

Viewed as a complex adaptive system the evolutionary process is artificially slowed if we stop experimenting. If we stop experimenting then it will take longer for us to learn the optimal algorithms, ways of doing things, and methods for self management/self regulation of this new space.

A problem with funding too many versions of the same experiment

One of the problems in the current blockchain tech community is that VCs and entrepreneurs seem to think very much in orthodoxy. A lot of money is flowing into duplications of the same experiments while some of the truly unique experiments don’t receive funding. As a result there are dozens of exchanges, wallets, and similar services which don’t actually innovate much except in marketing yet which are being flooded with cash. A likely result is a lot of these duplicates will fail, will be consolidated, and while this is good in some ways it is not good in all ways. Companies which truly bring innovative new ideas to the space should be given a chance to monetize them and build out but at the same time a lot of businesses are being built just to make money rather than to improve on how something is being done.


Snowden, D. (2010). Safe-fail probes. www. cognitive-edge. com/method. php.


Snowden, D. J., & Boone, M. E. (2007). A leader’s framework for decision making. harvard business review, 85(11), 68.