Proposal Discussion Concern about cheating and P2P handling of NPC simulation

[Dejay]

The game is in early alpha.

Well, sure, but this is the time to speculate I'm not bashing ED I'm just quite curious a little bit concerned.

But in the "worst" case FD would need to moderate and have a "report cheat botting" function and investigate and ban cheating players from playing online. That is always possible, but also costly to do. And from my experience it never worked well in other online games, it's just a huge hassle, delays, investigation, falsely banned players etc.

But yeah there is no cause for panic. Worst case, banhammer the cheaters.

If we are speculating that a single master does the NPC processing causing this zombie. Maybe they should consider a shared mastering over multiple clients.
They could be doing something like state machine replication using Paxos

Interesting! Does this work realtime (without delays for consensus) if only one node has the newest info?

 

[pompomJuice]

I meant I was harsh.

You were fine.

 

[KickahasTier]

Interesting! Does this work realtime (without delays for consensus) if only one node has the newest info?

Well you really hit the nail on the head.

Google has developed a API called TrueTime, which they use for their internal database Spanner

If I have it straight, the basic concept is that every clock (node) knows how uncertain it's time keeping is. That uncertainty gets bigger the longer the node is out of contact with a master clock (if you graph uncertainty vs time you get a sawtooth). When time-stamping a transaction the node basically waits until it is certain it can give an 'in order' and accurate (real-time) time-stamp.

Doing this (waiting for uncertainty to pass) enhances the reliability of the transactions in general and makes the entire system more efficient. It sounds counter intuitive but I've been led to believe it works really well!.

 

[Saxxon]

I have read this entire thread and have not seen another reference to preventions so forgive if I missed another discussion.

Considering I will not be playing this except in the multiplayer environment security and anti-cheat measures are extremely important to me. There will be no online game if these actions can not be reasonably secured.

The discussion on these issues may be in a less public place which I completely understand but if there is any official response I would appreciate a link if possible.

I cannot make a purchase decision until I know the facts.

Thanks for your time.

 

[Adept]

Don't worry, I think I know the research paper that inspired Braben, and it's very solid.

Look up Byzantine Fault Tolerance.

 

[KickahasTier]

The discussion on these issues may be in a less public place which I completely understand but if there is any official response I would appreciate a link if possible.

I cannot make a purchase decision until I know the facts.

Thanks for your time.

Security through obscurity is no security at all, if it's good security there should be a open public discussion about it.... somewhere.

 

[KickahasTier]

Look up Byzantine Fault Tolerance.

Interesting stuff, distributed state machine replication, secured by chain of work proofs (like in bitcoins blockchain).

So if there were an attacker, trying to corrupt the chain of work proof. How do long do you reckon it takes, to isolate them from network participation?

What would the thresholds for noise, and latency be? short enough that on a fast network and groups it rejects a false work proof in at most seconds?.

In a large group, agreement between as little as 3 would keep things moving, but it could be raised to increase protection against groups of attackers.

If there are a small number of participants (like the edge case of 2) is it still secure?

Any chance we can see some of those network code designs ?