Schrödinger's Army
Next: Time dimension in defense
Schrödinger’s Army
How to organise physical defence is one of the questions in decentralised world.
A virtual army is easy to set up allowing anyone to start one. An army of bits and plastics. This is how it works:
3D printers and other distributed manufacturing tools allow anyone to produce drones rapidly when needed. Likewise, explosives can be produced from synthetic fuels storing solar energy for winter. There is no need to make these during peaceful time, just to have the designs, skills and plans available in distributed manner.
This army can then act as distributed denial of menace towards any aggressors by being able to launch a massive attack from all over. Each drone has small impact but combined large effect. Parts of such a fleet can act independently, learn about its environment. Some drones acting as scanners, others taking orchestration functions, some delivering payloads. It's easy to think of such a concept being coordinated centrally when the need arises, but central orchestration is easy to shut down. It must be a self-organising decentralised response.
In addition to drones, robotic tanks and other machines will be used on land in conflicts of the future as human reaction times are simply too slow.
But nothing of this army does need to exists in real world during peaceful times. It is Schrödinger's army that only appears when someone tries to shut it down.
The question naturally is how does a decentralised army know reliably that an attack is ongoing? Everything is communicated through networks and sources for information can be compromised or messages from reliable ones blocked. Nothing is absolutely reliable but aim is to be as reliable as possible and auto generate the army only when an attack is happening and to prevent adversary triggering this unnecessarily.
Let’s sketch something based on how things work on blockchain, where such integrations towards real world are done with mediator services called oracles.
An oracle can be made of three parts: client who wants to know, mediator and data sources that fetch the results and report back to mediator.
Both the mediators and the data sources are decentralised.
To increase accuracy, results of query are asked from multiple data sources and the final answer is a combination of the results received in time. The mediator maintains a reputation score based on how reliable the results have been in the past and the query is sent to a random set of data sources that met the quality criteria (i.e. every time the query goes out to different set of data sources).
There are also multiple independent mediator services and clients may opt to use any of them or even ask from multiple and do their own consolidation of results.
As minor point, if the client wants to keep their queries private (i.e., not reveal where they assume an attack is occurring as this may reveal too much for the adversary), they simply encrypt the request with the oracle’s public key. Only the oracle can then read it.
Not a perfect solution as adversary could run a reliable data provider until they need it to fail or they need to compromise data sources and mediators. But even in this case they need to compromise a large set of independent parties or run a large set of devious services and even then, only achieve partial success.
Such a defence also needs to be broken into independent regions that make their own decision. Should adversary gain access to one region, it has no impact to other regions.
For more information see the white papers of popular Oracle services like chainlink or API3. On the downside these are heavily laced with web3 terminology that is intimidating for outsiders.
An issue with this scheme is naturally, that how to prevent people from using it to destructive purposes? The underlying technologies (3d printing, chemistry) exists and can be used irrespective of the Schrödinger’s Army concept. As the threat is distributed, the protection must be also.
But the problem persist and as a giant leap sideways, I’ll leave out defining perfect solution to the gentle reader as a mid difficulty level exercise.
Automated Armies and Forces that Govern Commons
You can apply the Power Flower diagram to analyse the situation between nations. Clearly a world where lots of countries have autonomous robots that can kill is a more dangerous as the one where we live today as force can be used without risking of any human casualties on own side.
Democratic nations cannot stop because dictatorships would then continue alone succeeding ultimately at some point and they do not have any moral objections to use whatever means possible to gain power.
Looking at the forces at play: rewards for violation are big, some dictatorships are too large and powerful so they do not face any real cost for violations, monitoring is difficult, no shared norms. Luckily still limited number of participants but strong incentives for small and medium size countries to join the race as they are the first ones against who these weapons will be used. Everyone has skin in the game.
Breaking the vicious circle seems difficult, but the Schrödinger army is a cost-efficient way to build defence under these circumstances for decentralised communities as long as the virtual, non-existent force is modern enough to inflict pain on attacker.
Another dimension comes to mind once one thinks of chaotic systems. Robots act much faster than humans meaning that counteractions must be automated as well. This leads to a large body of lethal machines with the only purpose of causing maximal damage intertwined into a complex and chaotic system. Complex, chaotic systems tend to have emergent behavior, that is not obvious from the elements. In this system only parts will be under the control of any party and giving falsified information about own pre-war activities has large rewards.
Such a system clearly needs a Chaos Monkey or Chaos Gorilla to pound out unexpected behavior. This can be done virtually in simulations if the physical actuators of these robots are disabled but the robots are fed with coherent but unexpected input. We talked about such simulations earlier in chapter 10 Stock Exchange.
Or alternatively two armies are brought to an uninhabited area where they duke it out physically while chaos monkeys delay messages, add incorrect input and so on. The emerging behavior is then studied to find out flaws and to prevent own robots going berserk and to detect when opponent is having a pear-shaped day.