White Flag is a Mistake

TL;DR: White Flag gave IOTA a nice TPS boost. But it’s temporary, increases dependence on Coordinator, and is bad for research, adoption & Coordicide development.

What is White Flag?

Let’s quickly discuss what White Flag does, and how it relates to the IOTA/Tangle concept.

Simplified description:

• White Flag allows invalid (conflicting) transactions to remain in the DAG. The coordinator will approve transactions that are invalid (directly or referencing invalid ones). When a milestone is received, nodes take all the transactions since the last milestone, sort them (total order over DAG), and walk down the list of transactions. Transactions that conflict with a previous transaction are dropped.
• In other words: The Coordinator’s milestones can now approve anything, and it is up to the nodes to figure out which transactions are valid and which ones aren’t.
• This creates a 100% confirmation rate, because every transaction (valid or not) can now be confirmed. This also removes the need for reattach & promote, as there are no invalid branches anymore. It also simplifies GTTA, because every tip can be attached to, valid or not.
• But White Flag is a temporary thing: It can’t work with Coordicide, so it will be removed again (“valid for the pre-coordicide era”).

At first glance, this seems good: It looks like this change reduces the reliance on the coordinator - the coordinator doesn’t actually do anything anymore except just sign a random transaction. Seriously, any transaction that indirectly references the previous milestone, valid or not, will do.

Primary Drawbacks

Unfortunately, this also breaks virtually every assumption that people have made about the Tangle, all the way down to “one transaction approves two others”. Some of the direct implications are pointed out in the “Drawbacks” section of the RFC:

The ledger state is now only well-defined at milestones, meaning that we have to wait until each milestone is issued in order to confirm a spend;

In other words, a “transaction approving two others” doesn’t mean anything any more, because transactions can now approve conflicting / invalid transactions)

Everything that is seen is now part of the Tangle, including double-spend attempts, meaning that malicious data will now be saved as part of the consensus set of the Tangle;

In other words, just because something is “in the tangle” doesn’t mean it’s actually valid, which also means that…

To prove that a specific (non-milestone) transaction is valid, it is no longer sufficient to just provide the “path” to its confirming milestone, but instead all transactions in its past cone.

So you can’t do merkle-proofs-of-inclusion with this.

Implications for Researchers & Industry Adoption

• It makes the DAG-structure useless, and turns IOTA into a blockchain (a poor one, as each block includes invalid transactions).
  • Because the coordinator can approve invalid transactions, every other transaction can, too. I think this cannot be stressed enough: Any transactions can approve invalid transactions, without any repercussions.
  • This turns “one transaction approves two others” into “one transaction references two others”. The references have no meaning, no value. Thus the whole concept of “securing the network with cumulative PoW (or VDF)” is meaningless.
  • This takes IOTA further away from where “it wants to be”, it effectively makes “The Tangle” paper completely unrelated to IOTA.
• This makes it less suitable for researchers evaluating tangle properties. For example, they can’t work with merkle proofs, they can’t evaluate security properties (“PoW validates indirect aprovees”), can’t evaluate performance (e.g. effects of reattach & promote)
• For industry partners, all of the above holds too, and creates additional uncertainties. Chrysalis makes it harder for them to evaluate how the network will perform post-coordicide. The same holds for open source developers & the ecosystem community, of course. Also, nobody in the industry seems to actually require the TPS increase right now:
  • Hornet already boosted the TPS quite a bit over IRI - but mainnet is running at 15 TPS at the time of writing, well below both the pre- and post-WF limits, all spam.
  • Private tangles exist as options for higher-tps testing.
  • Industry partners are not going to start seriously evaluating IOTA until Chrysalis Pt. 2 is done, because it changes almost everything anyway.

White Flag makes Coordicide way harder

Coordicide will necessearily make the network slower, WF or not (because you’re turning a centralized process into a replicated one, it’s slower by definition. Also, added overhead from Mana, FPC, etc).

But removing White Flag will slow down Coordicide a lot more. If I remember correctly, Hornet gained a ~5x speedup with White Flag. And this will need to be undone. I would imagine a 5x slowdown to be a tough sell to the community.

The alternative would be for the developers to invest additional time and effort in making White Flag work somehow on the post-coordicide network, thus making Coordicide even more complex (and delayed) as it already is.

Either way, painful.

Summary & Chrysalis Phase 2

TL;DR (2): White Flag cost months of engineering time to build, there is no short-term requirements to do this, there are no long-term gains but rather disappointment on the horizon, slowed adoption/research and of course additional engineering work to remove it.

So why have Chrysalis pt 1 at all? Is it worth it for a few TPS hype tweets?

Independent of my opinion on Phase 1, I think the changes planned for Chrysalis Pt. 2 are good & important and will fix some of IOTAs more serious current problems. I also want to stress that I don’t consider this the fault of individual engineers. The actual design on it’s own and the implementation are fine. But I don’t think management should have prioritized this project over Phase 2 or Coordicide.

There is some irony here, in that I have previously joked that “all of CfBs ideas that distinguish IOTA from other cryptos are being undone with Chrysalis, except the DAG & feelessness”. Well, the DAG is effectively being removed with Chrysalis pt. 1, even though it’s not explicit. I’ll write an article on feeless some other time ;)