Inflation/issuance schedule

[lrettig]

Preliminary research work is done: https://nbviewer.jupyter.org/github/lrettig/monetary-base/blob/master/jupyter-notebooks/SMH.ipynb. To be presented on research call on 3/1.

[lrettig]

Question for research: is there a way to measure no. of unique smeshers as a KPI (in place of, or along with, growth in effective network age/total realized spacetime, to measure performance relative to baseline) without incentivizing whales to run many individual smeshers, i.e., to fragment their identities? Can the system be parameterized such that an individual smesher is not incentivized to do this (since the cost of generating, and continuously operating, another identity would be too high) but the network as a whole is incentivized to add more unique smeshers?

[lrettig]

Brief recap of the discussion in yesterday's R&D call:

• Why did Filecoin decide to use effective network time rather than another metric to measure performance vs. baseline?
• In answer to the question I posed above, no, Tal thinks it's unlikely that we can parameterize the protocol s.t. miners are individually incentivized not to create multiple identities, but the network as a whole is incentivized to have many identities.
• Using total, accumulated, realized spacetime (like Filecoin) as a proxy for network maturity, utility, or performance sounds reasonable, as most (all?) other metrics can be gamed
• However, we feel that this approach is probably too complex and we'd all prefer a simpler approach
• We debated the idea of simple, constant, linear issuance vs. an issuance curve (with, e.g., exponential decay and a half-life). Iddo proposes measuring issuance speed relative to Bitcoin. Others felt that Bitcoin isn't the right baseline since the world today is very different than the world when Bitcoin was launched, and that it would make more sense to compare to networks like Chia and Filecoin.
• I feel that constant linear issuance is nice because, in addition to being very simple, easy to code, easy to understand and message, and perfectly predictable, it also lets the market answer the question of "how much should we be paying for security?" (via price discovery and the unit economics for smeshers)
• Tal also likes it because it's smooth and avoids phase transitions.
• We should plot constant linear issuance against the exponential decay function, but this isn't exactly apples to apples because the denominator (total supply) is infinite. We can plot cumulative issuance against total issuance at some point in time in the future instead, e.g., 30 years from now.
• However, we're targeting a very different sort of miner than networks like Bitcoin, Ethereum, Chia, and Filecoin. In particular, the small "home" smeshers that we're primarily targeting are probably a lot less concerned with unit economics, or even with price at all. In this sense, while we are competing for attention, we're not really competing for large, professional mining farm resources the way other networks are.
• We also discussed the challenges of governance and how hard it would be to change issuance in the future, especially if, e.g., miners have a say.

[lrettig]

See https://spacemesh.io/blog/spacemesh-economics-intro/ for the outcome.