What if Polkadot had 10% deflation instead of inflation?

DOT is an inflationary token by design. Its 10% yearly supply inflation has the goal to punish passive holders that don’t stake and imparts opportunity costs on crowdloan participants. Both staking and crowdloans come with a lockup period for the tokens (28 days in staking, up to roughly 2 years in crowdloans) – so the inflation could also be seen as a cost for having liquid tokens. Inflation is distributed between stakers and the treasury (image credit to @filippoweb3):

grafik

Inflation being 10% seems great for bootstrapping the network and is roughly in line with other chains in their early stages (BTC from 2013-2016, ETH from 2016-2018). As the network matures, we might want to reduce it. So I wondered – what would happen if we set this parameter at -10% instead?

The necessary changes would be that passively held DOT or DOT contributed to crowdloans “melt” at -10% p.a., while staked DOT don’t (at the ideal staking rate and minus a commission rate for nominators as is the case today). The treasury would be treated as it is now – meaning, at the ideal staking rate, its funds melt at the same rate as a normal holder, and more slowly above/below. Its income would come from transaction fees and slashes (and perhaps coretime sales in the future).

Economic network security considerations would also change. Each DOT would, ceteris paribus, become more valuable in its purchasing power as the total supply decreases over time. Loss aversion might cause more people to stake their DOT versus passively losing through inflation. A real-life example of this would be the different reaction from the public when banks in some European countries charged an actual negative interest rate to EUR savers (around 2021) compared to when “only” real rates were negative (e.g. interest rates at 0% and inflation at >0%). The liquidity imbalance would also shift, as there are less natural/forced sellers of DOT – for example, through the obligation to pay income taxes on staking rewards. How such a system would be treated tax-wise depends on local tax laws, but no staking rewards → no staking income tax seems logical.

How this would work out if actually put into practice, I don’t know – and I’m sure I’ve missed quite some nuances that would make implementing this tricky or unfeasible altogether. However, it is not that different from the situation of today, only that certain aspects are less hidden: Non-stakers and crowdloan participants already implicitly pay for the staking rewards of stakers through inflation, and all DOT holders collectively pay for inflation spent by the treasury.

This is mostly meant as a thought experiment and I would welcome all contributions on why this would be a horrible or great idea. I think that when discussing changes to Polkadot’s inflation policy, even more gradual reductions or increases, it helps to consider rather extreme cases to visualize how each stakeholder would be incentivized to act and how incentives change as the inflation parameter shifts.

5 Likes

This is actually a discussion that I have had in the past.

Generally speaking, what you are describing with all passive holders loosing 10%, rather than all active stakers earning 10% is economically the same, and a deflationary model is certainly preferable for a number of reasons, not least of which would be reporting of taxes / income. So you are certainly coming from a place which makes sense.

However, from a technical perspective, designing such a system is much more complex and has much more overhead.

For example: All changes on change need to be triggered by some internal or external action. We cannot have Polkadot automatically decrease the balance of all passive users (which is a large majority of the token holders). Such an action would be extremely resource heavy. Imagine the on-chain costs of updating 1M users account balances, and how that will increase as more DOT holders come.

Fortunately, for staking with rewards, a small minority of stakers are already incentivized to make on-chain transactions to claim their rewards, paying for and triggering all the on-chain logic.

Also imagine from the user experience perspective: you transferred some balance to a cold wallet, and then a year later you come back and 10% of it is gone, without you taking any action. Imagine you start with just 1 DOT, then you go below the existential deposit, and your account gets deleted!

Finally, the inflationary model is one which supports ideas like the treasury. As your graph shows, some amount of the inflation goes to the treasury to be spent on improving the network. Discussion here is actually trying to discuss how to ensure that treasury has a consistent income from inflation.

With your idea, I guess we could simply move funds from users into the treasury account, and have that be the “deflationary” pressure, but again, such a process unfortunately just doesn’t scale.

However, that is not the end of the discussion. While it is not practical to me to create a deflationary system, there are other kinds of economic primitives which can provide similar kinds of behaviors as a deflationary model, for example Non-Interactive Staking.

In summary, I think the inflation system implemented today is basically a deflationary system to passive token holders. I think it is just all perspective, and the underlying economics is mostly the same. I think it would not be crazy for Polkadot to have used the deflationary model, except that it is technically non-feasible.

4 Likes

An interesting thought experiment indeed, good to have these prompts since it does help to unlock certain topics at a very core level.

Under the right conditions / context this could actually make sense to encourage spending / velocity of money were that desired, it’s basically demurrage. In many cultures interest on money is labelled ursury and forbidden. It’s also become a focal point of the “degrowth movement”.

Demurrage is the cost associated with owning or holding currency over a given period. It is sometimes referred to as a carrying cost of money. For commodity money such as gold, demurrage is the cost of storing and securing the gold. For paper currency, it can take the form of a periodic tax, such as a stamp tax, on currency holdings. Demurrage is sometimes cited as economically advantageous, usually in the context of complementary currencysystems. - Wikipedia

It’s a very interesting area and there’s been a small amount of experimentation but it would def be an area to play around with - we’ve discussed it a fair bit over the years and in relation to the economic design of Kabocha.

See Freicoin which is a real crypto OG.

If you follow the demurrage thinking and managed to turn Polkadot into an active fee generating economy - by encouraging spending, rather than compounding/hoarding, you can still have a treasury sustained not by inflation but by economic activity. Now there’s a thought…

Hi Shawn,

Thank you for sharing your thoughts and technical insights.

Your explanation regarding the technical infeasibility (and UX) makes total sense, I see how this would be too resource heavy. I assume that this would be the case even with inflation at 0%, where stakers earn (they pay for and trigger on-chain logic) and passive holders lose, since each reward claim would still come with way too many balances to update? Unless there would be a way to decrease specific balances only (e.g. matching DOT amounts), which also sounds impossible.

If I understand correctly, that would mean we are technically limited to re-scaling the y-axis in the inflation graph rather than shifting it downwards, to prevent any negative rates that lead to continuous balance decreases.

I hope that in the long run, the treasury can fund its operations sustainably through a non-inflationary source, be that transaction fees or coretime sales or something completely different. Non-interactive staking also reminded me a bit of the idea of an “on-chain bond” (mentioned here) - if, instead of being locked up, the DOT are loaned to the treasury and this is reimbursed by compensating for inflation plus a premium, it could present an alternative way to bridge liquidity until the (non-inflationary) income surpasses spending at a later stage in the growth cycle.

You referred to multiple other economic primitives to mimic the deflationary models - are there any specific ones on your mind besides non-interactive staking?

Hi Rich,

Thanks for bringing this up. I think the comparison to money may not be appropriate - DOT is software and a way to access Polkadot’s services. Many problems associated with monetary deflation in real-world economies, such as increasing the burden of debt, would not exist, as there is currently no DOT-denominated debt. It might incentivize some hoarding behavior, though, as each (staked) DOT unlocks slightly more blockspace as time goes by and the total supply decreases.

2 Likes

DOT is “software” is a useful legal narrative, but that is a very different thing to the reality of the role DOT has in practice.

DOT is attempting to be money as it is at the core economic asset of a sovereign network economy. If it weren’t sovereign then the Web3 proposition would be moot - which it may well be - only time will tell.

In this sovereign network economy, DOT is attempting to fulfil the three principle characteristics of money for a select collective of participants:

  • medium of exchange (contributors paid in DOT)
  • unit of account (services priced in DOT)
  • store of value (accrues long term value through the provision of MoE and UoA)

Try as the Polkadot / Parity team might, you cannot market and position Polkadot in the same way you might with other SaaS propositions - it’s impractical, myopic and detached from reality.

2 Likes

I think if you gave someone the challenge of implementing deflation on a blockchain, they could be forced to create some systems which would work.

For example, imagine a system where each account on the blockchain has both a balance and a timestamp of the last time they made a transfer. Then, when doing a balance transfer, you do some math which calculates the “deflation” on that token since the last transfer.

For example: I have 100 DOT which was last transferred 1 year ago. If I want to send my funds to an exchange, I call transfer(100) or perhaps transfer_all(), but the recipient account only receives 90 DOT, and the other 10 is cut due to inflation, perhaps sent to the treasury.

The thing is, these kinds of solutions are always compromises, and open up many non-ideal paths. For example, perhaps you now encourage trading private keys rather than the tokens. Perhaps you discourage use of the token at all! I could create many such systems, all of which have some kind of problem, just as inflation may have its own problem.

The funny thing about humans is that they have many cognitive biases which lead to different behaviors even if the situation is technically the same. I think one nice thing about Substrate is that game theorists and economists can quickly launch their own blockchain with their own ideal tokenomics and see how it goes.

1 Like

Yes, deflation encourages hoarding and discourages transactions and economic activity

the magic of the system indeed.

I’m not sure that implementation is difficult: an implemntation can be a view on another regular token with a factor which is decreasing:

Let’s name our currency DEFL,
We have a global “factor” as a storage value: F initialised as 1
When handling DEFL on chain we handle in the form of another token, let’s name it TOK.
TOK is implemented as a regular token.
The number of DEFL for account A is defined as such: "number of TOK for A" * F
If we want to generate a reward of 1% of the DEFL tokens by decreasing everybody’s balance we do: F = F*0.99 and then create "total issuance of TOK" * (1/0.99 - 1) TOK
(if my math are correct)
If we want to print new DEFL (using inflation) we do: create new TOK and don't touch F

For existential deposit I feel it is quite the same as a regular token

As crypto is a software, what we define as the number of DEFL for an account is just a view on the blockchain state.
In the example above TOK is just an implementation detail.

But all in all I’m not sure if this changes anything. Because the value of the token is related to usage and prediction on future usage, the price of the token relative to other token in the exchange market shouldn’t change so much.
About tax: if tax is about the price of the token in euro (or other) at the moment of the exchange then it shouldn’t change too.

2 Likes

@gui in this case, what token is a user holding?

What token is used on an exchange?

I do see that a simple conversion factor between two tokens could convert one inflationary token into another delfationary token, but it is not clear to me how we force users to hold one over the other.

And if there is some underlying conversion factor in the background, and only one underlying balance, then it seems like it is just a UI thing.

For example, we could simply create a “deflationary token” with DOT by having UI’s use the following math:

shown balance = (on-chain balance / total-issuance) * dot-fixed-supply

or something similar to this.

However, this would not change the fact that there is some simple underlying token which is inflating.

@gui in this case, what token is a user holding?

What token is used on an exchange?

User holds DEFL, only DEFL goes through exchange, the other token is an implementation detail.

And if there is some underlying conversion factor in the background, and only one underlying balance, then it seems like it is just a UI thing.

I disagree, what makes account A having X DOT is not because there is some substrate storage map which associate A to X. The reason we consider A having X DOT is because there an agreed reliable way to evaluate how many DOT have A, and the ability to transfer it.

In our case TOK is just an implementation detail, we don’t even need to give it a name in the actual implementation, we can just not name it.
Everything is a UI thing. Like when we did redenomination the actual value stored in the substrate storage map hasn’t change, it is just a UI thing.

1 Like

There is one with extrem deflation, which is BNB. Deflationary maintain the token price for the short term but discourage ecosystem for the long term. The system itself is easy to implement but not good for the long term.

1 Like