Dynamic plasma discussions

Development
81

I’m perfectly fine with this compromise. In fact, if we learn from how other ecosystems evolve i.e Ethereum, we can comfortably say that whatever we end up implementing now will be disregarded and changed in the future as the technology matures and better solutions are found.

I think we can all appreciate the effort and understand that a perfect algorithm for transaction ordering that is battle-tested, peer-reviewed, with a solid game theory/economic foundation that will stand the test of time is probably unrealistic at this point.

That said, there are still a lot of other devs who have not voiced their opinions, so I’m very eagerly awaiting their take, as clearly the implementation of Moonbaze’s proposal or suggestions for adaptations to it’s proposed solution is what matters now.

82

summary of my thoughts:

  1. QSR is a utility token to launch pillars, sentinels and process transactions.
  2. QSR that is sitting idle (not burned for Pillar or staked for Sentinel) wants to be used. Either it can be fused for transactions or leased to others for transactions (a market that will form in the future). There is an opportunity cost to fusing QSR. A spammer will consider these costs when determining if they want to use QSR to spam the network. Will someone attack the network with fused QSR? Maybe, but there is a cost to that.
  3. Can someone attack the network with PoW? yes. How can we mitigate it? Can we establish the “lanes” at the block level? Can we limit TXs in the block / momentum based on PoW or QSR?
  4. I hate to bring up guidance from Kaine, but it’s hard to avoid. Given my lack of experience with protocol development I need to rely on what he has said in TG. If Kaine envisioned a fee to process TXs he would have mentioned them. Would the early devs have marketed the project as feeless if we planned to implement a fee in the future?

Screenshot 2023-09-23 at 5.16.30 AM
Screenshot 2023-09-23 at 5.16.30 AM1422×310 41.4 KB

  1. I can see how fees will help order congestion and provide a fast lane. That makes sense to me. It’s hard to ignore George’s guidance that in the future it will be hard to differentiate between spam and network usage. And if we have a fee lane, why won’t all TXs migrate to higher and higher fees?
  2. Do we think Kaine will issue a spork ID for fee lane given his guidance never mentioned a fee.
  3. How do PTLCs impact this discussion? Given this will increase privacy will it become harder to differentiate between spam and honest traffic? Won’t everything just move to valid TX or not?

If we can limit TX in a block based on PoW / QSR can’t we implement lanes? And if we believe there are opportunity costs to fuse QSR isn’t that the fee we are looking for?

83

On a separate note, George pointed out that without Virtual Voting, there is nothing that requires the Pillars to follow the plasma ordering we are discussing. That lead me down the “Virtual Voting” rabbit hole in the whitepaper with ChatGPT. Here is what it says about our virtual voting.

This is off topic and not relevant to the discussion about dynamic plasma.

Consensus Timeline & Virtual Epochs:

The consensus timeline is divided into virtual epochs. Within each epoch, nodes process transactions and make decisions based on the total stake present during that epoch.

Voting Mechanism:

The node will broadcast its vote in the next epoch.

If a node knows the transactions between two epochs and applies a deterministic ordering algorithm, all remaining honest nodes will arrive at the same decision. After a node has a supermajority of messages with all transactions between two epochs, it starts to virtually vote on the ordering.

This virtual vote isn’t a traditional vote. Instead of sending additional network messages, it uses a set of rules that define a deterministic way to order the transactions. Factors used in ordering include the PoW link weight, the timestamp of transaction arrival, and the hash of the transaction as a tiebreaker.

Once a node orders the transactions, it knows that the order is the same for all nodes. Thus, it marks them in the ledger, assigning an ID to each transaction.

Stake-Based Voting:

In a decentralized environment, traditional quorum-based voting is vulnerable to Sybil attacks. To counteract this, nodes can lock a certain amount of stake to obtain different roles in the network, such as becoming sentinel or pillar nodes.

The virtual voting process is determined based on the total stake during a virtual epoch. Pillar nodes with stake make decisions within the consensus algorithm to finalize transactions. Nodes can unlock their stake at any time. However, consensus nodes have an “unstaking period” they must wait through.

84

Something we didn’t discuss (or I missed it) is the way the NoM architecture might or might not react as a blockchain in case of spam or congestion. I think Sol tried to spam the network once but I’m not sure we have any other metrics.

On a side note please be more careful with tools like ChatGPT, it’s not magical, and it hallucinates a lot, like a lot. It should only be used when people actually know the content that is sumarized or the subject they ask questions for, so they can check the facts by themselves.

85

I independently verified the general accuracy of that before posting it. I agree with your caution however. It does produce random and wrong results often.

2 Likes
86

I’ve been following the KAS dust attack. I think basically they are censoring TXs that have the footprint of spam. They seem to be successful so far. Someone claimed it was a miner looking for higher fees to mine blocks. They got censorship instead.

87

If you need to censor it looks like your system is not that good.

88

More ideas are coming to mind. I’m going to vomit them here.

BTC has a difficulty adjustment so blocks are processed every 10 minutes. THORChain has an incentive pendulum to balance (adjust) the amount of Bond a node earns vs the LP incentives. If there is too much node bond, node yield goes down and LP rewards go up. This reduced node bond and increases LP depth. If there is not enough node bond, node yield goes up and LP rewards go down. This increases node bond and decreases LP depth.

Why can’t we have a difficulty adjustment on PoW and QSR to balance the depth of the queue per lane. The goal is to balance the TX across the two lanes. For example:

  • if the PoW lane is congested with long wait times, lower the difficulty on the QSR lane and increase difficulty on PoW lane to incentivize traffic to the QSR lane.
  • if the PoW lane is wide open with short wait time and QSR is backed up, increase the difficulty on the QSR lane and lower difficulty on the PoW lane to incentivize traffic to move to PoW.
2 Likes
89

Individual sovereignty and censorship resistance of the network can co-exist.
In a system where validators/miners have choice over which transactions to confirm, they will have the ability to censor.
The question becomes are there enough competing interests in the network that are unable to collude to censor. That’s a big component of decentralization.
And where censorship resistance would come from.

Other mechanisms like threshold encryption and virtual voting can greatly increase the censorship resistance of the network. But they remove aspects such as choice.

1 Like
90

Don’t forget about this post. It explains how PoW and Plasma work today.

91

We don’t know if Mr Kaine v2.0 is the real one. I would not trust a stranger so much. He didn’t even know that PoW does not secure the ledger. I think in the end we would all reach the same conclusion, optional fees is a good solution for today’s problem. In a couple of years if we grow bigger, we could update the model, with better solutions, if any.

92

This kind of speculation shouldn’t even be taken into consideration. It’s near complotism level. At this point, we don’t even know if Moonbaze is the real one.

93

Again, you are just trolling the conversation without offering something useful to reach consensus for what solution we should choose.
Off topic: tell me what percentage to change for delegation and I can prove who I am.

94

When Kaine snapped back at the community a few months ago and started to talk a little trash, I renamed him (in my TG contacts) to v2.0. It’s possible the person behind that account changed, but it’s the same account that has been posting for years.

I agree that we need to do what is best for the network. I view the Kaine posts as important insight we should consider. Nothing more.

95

For what it’s worth, the dust attack on $KAS is still happening.

Screenshot 2023-09-28 at 9.13.07 AM
Screenshot 2023-09-28 at 9.13.07 AM1544×370 45.6 KB

96

Not sure if this is relevant to our discussion regarding plasma. Just wanted to share some of the consequences of this dust attack.

Screenshot 2023-09-28 at 9.15.05 AM
Screenshot 2023-09-28 at 9.15.05 AM2324×672 136 KB

97

Why can’t we implement a plasma pendulum that tries to balance PoW with QSR? We can limit PoW attacks by limiting the number of TXs per block that are based on PoW. And if we create an incentive mechanism to move TXs from PoW <> QSR based on congestion, why can’t that work?

Also, I do think we need to consider the opportunity cost of fusing QSR as a fee. If a user wants to make sure a TX gets processed, fuse more QSR. Once a spammer’s PoW gets throttled, they will need to post a lot of QSR to spam the network and there is an opportunity cost of that.

98

I’m having a few offline conversations, and wanted to throw out this idea based on conversations w/ others.

Do we expect to onboard all new users to the L1? Is that a design objective? Kaine has told us for months / years that he wants to keep the L1 clean and minimal. I assume most user activity will happen on the side chain and/or L2. What will new users do on the L1? Stake, delegate, get rewards, launch sentinels, and eventually pillars. What else will new users do on the L1?

So if we assume most users will not need to interact with the L1 frequently how does that change our design decisions?

In order to Stake and Delegate do users need to interact with the L1? If PoW is under heavy load, how does a new user perform PoW to receive QSR? Can users do that from the side chain or L2? How can we insure new users can perform enough PoW to perform their first few TXs to get QSR? Once they get QSR they can hopefully avoid a PoW bottleneck.

2 Likes
99

Yeah i agree its the dynamic aspect of it that seems to have potential.

Balancing between lanes and some sort of “difficulty adjusted” floating base plasma if possible.

100

If we assume most users will not interact with the L1 ever or infrequently, this might change our mindset.

And if a user needed to interact with the L1 why can’t they run their computer and built up PoW “credits” over time. If they don’t have QSR and need to make a TX on L1 and they only have PoW let them run their computer for some time to build up cumulative PoW.

Another option is the user pays a 3rd party for PoW to process a TX for them. 3rd party fee markets will spring up to generate PoW.

Another option is to have some sort of springing transaction where a user pays a fee on the L2 that once conditions are met, QSR is fused on the L1 for the user allowing a TX on the L1. QSR owners could rent their QSR to users for L1 TXs.

I’m starting to think we should not be designing systems that allows “regular” users to easily interact with the L1. Today that is all we have so it makes sense we are focused on that. But over I think we plan to migrate activity to higher levels.

2 Likes