Why did we need multiple addresses ?

[aramakam3505]

In comsos for a validator we have 3 types of addresses - AccAddress, ValAddress, ConsAddress.

What was the consideration behind maintaining 3 types addresses for a single validator?

[alexanderbez]

The original backstory to this is the following:

You need to understand, from a validator's POV, there are two distinct key pairs -- the CometBFT key pair, which they use in consensus and a application domain "operator" key pair, which they use for signing on-chain txs.

The ConsAddress is a bech32 HRP for the former. The ValAddress is a Bech32 HRP for the latter.

[alexanderbez]

I think you misunderstand. It's Bech32 HRP that's different. The underlying key is the same for AccAddress and ValAddress.

The consensus key pair indeed does use a different curve for a multitude of reasons. Even if the consensus key and app/state-machine key used the same curve, you wouldn't want them to be the same for many reasons (e.g. security)

[MindlessSteel]

This makes perfect sense too me .

[teddyknox]

@alexanderbez Can you elaborate on that last bit? Why does the consensus key pair use a different curve? Why wouldn't it be ok for them to be the same key if they were the same curve?

[robert-zaremba]

@teddyknox it's a design choice. Technically they could use the same scheme. For consensus ed25519 is used - it's faster to verify (and has other nice properties). App keys are same as Ethereum / Bitcoin - because of wider adoption.

[alexanderbez]

Yes, what @robert-zaremba said is correct. Again, even if they were the same curve type, you wouldn't want to use the same key for both consensus and app operations because that would be security/opsec disaster waiting to happen, e.g. hot wallet keys getting hijacked. You don't want consensus critical keys being used for everyday ops like voting and transfers.