How do you provide trust in the network?

[DeveloperAlly]

In Lilypad, how do I know my compute job will be run as intended by a Compute Provider, and I won't just get back a random result?

ie. if I made a docker container that computes 2+x could a node just return 3 no matter what the value input?

[DeveloperAlly]

Lilypad approaches compute verification using a variant of optimistic with economic collateral staking => this is best for performance while ensuring trust in jobs with game theory

What this means in practice is that there are mediators running on the network that check a % of jobs to ensure that jobs are run correctly AND compute providers must provide collateral (at ~10x the job amount currenlty) to win a job. So, if a compute node does not run a job correctly and it is caught, it's collateral is not returned. A client is also able to challenge results (for a refundable fee).

This set-up operates much like validators operate in L2 networks - currently these mediator nodes are run by trusted parties (the Lilypad team), however in future, this will become a more decentralised part of the network and be a consortium of nodes.

An easy analogy I like to use is that of ticket inspectors on a train - if the price of the ticket is low enough and the cost of getting caught high enough, alongside the probability of being caught being high enough - then you will probably do the right thing and buy a ticket (run the job faithfully).

A great starting point for deeper understanding is found in the docs here: https://docs.lilypad.tech/lilypad/lilypad-aurora-reference/game-theory