In 2018 I started forming an idea based on the CAP theorem of distributed systems theory, that one can get all three features, consistency, availability and partition resistance, only if you have several cooperating and interacting protocols running in parallel, independently, but intertwined.

My initial idea put PoW and PoS federation style alongside a fast gossip based PubSub - I still think that the Mempool could become its own eventually consistent protocol and strengthen finality and reduce reorg risk without centralisation.

Obviously, as a pure Maxi now, the federation of any form is a liability, it needs to operating on the peer to peer level as a baseline.

But state channels, now, state channels, plus the proof of work blockchain, plus the fast gossip pubsub consensus filter...

Indra easily fits the role of a transport for the mempool system, in fact, it makes a lot of sense that p2p traffic be anonymised and the majority of it running out of hidden services.

The security budget for a global monetary system is pretty big. The conventional system is much higher and more risky due to the often old and antiquated methods of securing the connections. The wide adoption of visa/mastercard based payment networks is exposing a great deal of vulnerabilities that sooner or later will be recognised and a lot of people suddenly without any money.

So it makes sense that if I'm in the business of sending out sats to pay for things, from a secure, large stack, that ideally, nobody knows where the keys are.

For generic traffic, just gossiping things, this is not relevant, but considering the two things together, how to make the gossip help hide the traffic, you put it into the same layer.

Because this control traffic is metered, but privately measured, it's a bit hard to see but if nodes pay for their more sensitive outbound traffic, and all of them do it, then that percentage of their income is going to literally be reciprocal to their spend.

The blockchain, with proof of work and probabalistic finality, combined with state channels that work within that static, expensive to change global state, to facilitate routing payments, obviously there is one ingredient missing from all this:

Hiding where the signals emanate.

I believe an analysis of the C, A and P properties of these three patterns of system design, they overlap and produce a strong element in each case.

Bitcoin's *consistency* is the legend of it. It is not immediate finality, but after 6 blocks it is infeasibly small odds of the history being changed.

Lightning's *partition resistance* is what makes it work (It is source routed but no peer needs to know the whole network, only paths and the onions give addresses), and generally the burden is on the merchant. Literally, channels cannot be deleted, and they continue to exist at least in potentia while someone has the keys to the transaction and the latest state of the channel. It is a subtle distinction in some ways but in others it's very clear - the state is locked by bitcoin's consistency.

Indra protects *availability*, the primary threat against availability is of course the lack of paths between two points. In network terms this is firewalls, blacklists, jamming/shutting down satellites/base stations, and cutting cables. This is also why Indra's most important type of control messaging is peer metadata.

As a trio, all three reinforce and create the utility for each other, providing privacy, store of value, payment processing, and privacy of communications, all being essential tools for business, and implemented so that the cost of compromising the security of the networks is beyond the potential loot payload.