[Brainstorm] How do we ensure small vaults have a place at Maker?

This post is intended to offer a broader discussion on strategies MakerDAO members see fit for maintaining the ability for smaller vaults to exist. You may want to check out the conversation on the Bite Rebate Poll to inform the discussion here. The Signal Request to Increase the Dust Parameter is expected to be finalized by on-chain voting next week, raising the Dust Parameter (minimum DAI debt per vault) to 2,000 DAI.

The security of the protocol is paramount, which helps explain why the Dust Parameter was just 20 DAI in August of last year and will now (probably) be 2,000 DAI just five months later. If we do not put together a plan that allows for liquidations to be preformed more cheaply, our Community will be at the mercy of Gas prices and ETH appreciation, preventing many valuable operations from people not currently in our community.

I will use one last wall of text here to briefly explain why this issue is so near and dear to me. Put simply, I would not be here contributing to this vibrant and supportive community if I had found out about Maker even a couple weeks later than I did. For me, being able to test the vaults by borrowing the minimum (at the time) of 100 DAI was paramount for me being able to understand and trust the system with more of my assets. Less than a week after opening my testing vaults, the Dust Paramater was increased to 500 DAI, prompting me to join the community and start posting on the forum. I am constantly grateful that I found MakerDAO mere days before it might have slipped away from me, as I now consider this community a home and a welcome personal obsession.

Seeing as the dangers of a higher Dust Parameter include:

  • inability for new users to “test” vaults at reasonable risk levels
  • a barrier to entry in the DAO for contributors with less capital
  • effectively pricing out users with a native currency weak to USD
  • diverting market share to platforms (like Aave and Compound) without high minimums
  • more concentrated positions from individual vault users
  • making compatible “partner” solutions like B.Protocol more difficult to manage

And a variety of different vectors for solutions have been suggested like:

  • rebating keepers that call bite on undercollateralized vaults
  • incorporating a Layer 2 solution to cut down on expenses
  • expanding partnerships with groups like B.Protocol that offer committed liquidators
  • updating Liquidations2.0 to be more efficient for liquidating smaller vaults
  • moving small undercollateralized vaults to an ETH reserve fund
  • hiring someone to manually preform bite calls

I think it’s time we start aggregating and discussing what might be a reasonable path forward.

There are reasonable concerns about bandwidth and development cost, but I firmly believe the amount of vault users and contributors to MakerDAO we will gain in the long run by investing into a solution will cover those costs ten-fold.

Thanks for reading and please drop your thoughts below! If you like an idea you’ve already seen in the forum, I suggest giving a tag or a linkback to draw in some discussion!


or alternatively/additionally give higher (relative) liquidation penalty to small vaults.
if a minimum liquidation penalty would be 50 to 200 DAI + 13% of the debt, then the LIQ 2.0 system would safely be able to give 200 DAI tip for the caller of bite.
for vaults with 150% CR, it would probably allow supporting minimum debt sizes of 200-1000 DAI (depending on the actual tip, and the efficiency of liq 2.0 process)


I wonder if having a higher LR would allow smaller dust as well… But cannot get the math around it done :wink:


This is something that can likely be done once we have a viable L2 solution such as Optimism. Until then we are stuck with the heavily congested main chain. I don’t think there is a way around it.


Wow way more initial responses than I expected on a weekend lol thanks everyone!

I like this idea a lot and it wouldn’t be too hard to implement. The goal after all is not to get liquidated and even if someone wanted to test liquidations, having a low Dust limit with a higher penalty would seem agreeable.

Likewise for this idea! If the dust is low enough, being forced to open with more collateral would still be way cheaper than a higher dust limit. What if we added a high LR, low dust vault arrangement for each new collateral type?

Hearing xDAI present on this week’s collateral call definitely makes me optimistic that we could implement some L2 solutions imminently. Out of curiosity, are we currently running any experiments with L2 compatibility? I know @NikKunkel mentioned that an L2 oracle is desired on the last Governance and Risk call, seems like that would have to be the first step in putting something together

What if
We keep the dust parameter.
But we add a new parameter to switch from AMM vs dutch auction.

Under x amount, auto liquidation via AMM at a swap cost.
Swap should be the cheapest options to liquidate.

Over x back to doutch auction as we currently do.

Amount under the attack vector can be calculated if we own the liquidity and know the swap fees.

For the actual kick of the auction either we incitive people to start the auction via some reward or we loop on those specific vault during the price changes.

Also on another hand the attack describe is not really applicable as the attaquant needs to show it is game way before the attack. And it is visible we would pass from 20k vaults to 1M vaults. At that time if we increase the dust the attaquant is the attacked as he needs to pay a lot of gas to exist its position.

Or I missed something but this attack needs a lot of capital with a big chance of losing it.

Actually, would taking a position on gas saving token more profitable.


Totally agree–we will be stuck w/L1 for the foreseeable future–you got Compound announce their App-chain, we got Layer 2’s coming online and they won’t be directly interoperable and composable the way Layer 1 is. I myself tried Loopring and left because there was barely any liquidity. All in all we are going to have to think about Layer 2 DApps a lot different than the way we think about the basic Apps we use today.

I actually came across this conversation this week:


That is the solutions I prefer overall, eth and wbtc should be enough for new joiner. And as said actually eth-c is not really interesting except for rate psychological reason.

In addition, one of the pain points for small vaults when raising dust parameter is also how it impacts existing small vaults. It’s my understanding that small vault holders have to lock up additional collateral just to mint big enough vault and then close it instead of closing their small vaults directly. Correct me if I am wrong though.


no. in fact they cannot lock up additional collateral before they would first either close their debt, or increase it (and the latter might not be possible if their collateral is small).

1 Like

Ah ok. Thanks for clarification.

1 Like

Erm… I guess that is not correct, you should always be able to add more collateral.

I guess what you mean that it might be complicated to release the necessary collateral to swap it for DAI to repay and with that closing the vault. Repay with flashloan would help here, I heard defisafer is already offering something like that

see this line in the code.
it checks that debt > dust, and revert otherwise, even if the operation is ilk deposit.

I was also surprised/disappointed to find it our, when one of B.Protocol users was not able to deposit more collateral after the last dust level increase.


Oh… wow… that was unexpected. On the other hand: if you have more collateral to add you can easily swap to DAI to repay and with that closing. It is a lot harder if you have nothing to add - then you are basically locked (without additional tools)

Only if your would be added collateral is entirely enough to repay the loan though. You can’t even pay down DAI debt when your vault is below the dust minimum unless you are paying it off entirely.


Unless the collateral is not enough to repay the full debt, but it is enough to save save you from liquidation.
Then, sfyl.
A flash loan can save u here, but it force u to trust other systems.

1 Like

DAI native flashloan is already near (weeks not months), so could have build something native


I think the answer here is for a “mega-vault” solution. A protocol (maybe @sowmay @Samyak with Instadapp or @yaronvel with bprotocol?) could open one big vault and then manage the users positions (and handle liquidations) external to Maker. The only job of this sub-protocol operator would be to ensure that their own liquidation system worked well enough to ensure that the “mega-vault” is never liquidated.


Our proposal for committed keeper mechanism

could also support such low dust level, as instead of liquidating the vaults, it just add unsafe vault position to the keeper vault, which is pretty light on gas.


Can you explain a bit further? I thought that if bprotocol failed to liquidate a vault, it simply reverted to the normal liquidation mechanism? Are you saying your implementation would batch small vaults into a larger position?