PI-3: Bug Fixes and Parameter Changes

Executed on December 5, 2024 at 02:52 UTC.

Summary

1. Rain Stalk Preservation: Update the Silo such that Deposits do not lose Rain Stalk upon transfer or Convert into non-Germinating Deposits.

2. AL2L Converts: Re-add Anti Lambda to Lambda (AL2L) Converts.

3. gm Reward: Update the gm reward to properly issue Pinto when the gm function is called 239 to 240 seconds late.

4. Flood Pods: Update Soil issuance above peg to properly account for Pods that became Harvestable due to a Flood.

5. Soil Issuance Above Peg: Change the Soil issuance above peg multiplier on Pods that became Harvestable from 1.5x to 1.2x when the Pod Rate is Excessively Low. Stop halving above-peg Soil issuance when the Pod Rate is Reasonably High and Pinto’s price is above the excessive-price threshold. Retain the 0.5x multiplier for Excessively High Pod Rates.

6. Morning Auction Duration: Extend the duration of the Morning from 5 to 10 minutes.

7. Minimum Temperature: Change the minimum Temperature at the beginning of the Morning from a fixed 1% to 1% of the Maximum Temperature.

8. Maximum Temperature Increases: Reduce the Maximum Temperature increase from 3 percentage points to 2 percentage points wherever it currently increases by 3 percentage points.

9. Excessively Low Pod Rate Threshold: Lower the Excessively Low Pod Rate threshold from 5% to 3%.

Links

• Implementation PR

• Implementation Commit: 6f65b5ecbe83fe07cf48982963e0203379cfaf0e

• Safe Transaction

• Diamond Cut Transaction

Problems

1. Rain Stalk Preservation

Rain Stalk is the snapshot of Stalk holdings taken each time it begins to Rain, which determines pro rata ownership of the proceeds from the Flood sale. Rain Stalk was erroneously burned upon the transfer or Convert of a Deposit.

2. AL2L Converts

AL2L Converts were disabled in PI-2 due to a bug where all Rain Stalk was erroneously removed as a result of an AL2L Convert while it was Raining.

3. gm Reward

The gm reward was 0 Pinto when the gm function was called 239 to 240 seconds after it could have been called.

4. Flood Pods

Soil issuance above peg is intended to be a function of the number of Pods that became Harvestable at the beginning of the Season. Pods that became Harvestable due to a Flood were erroneously omitted from this calculation.

5. Soil Issuance Above Peg

Soil issuance above peg was too aggressive when the Pod Rate was Excessively Low. The 0.5x multiplier intended for Excessively High Pod Rates also applied when the Pod Rate was Reasonably High and Pinto’s price was above the excessive-price threshold. This incorrectly halved Soil issuance in those cases.

6. Morning Auction Duration

The previous 5 minute duration of the Morning did not give sufficient time to participants in the Morning Auction who are submitting transactions manually to participate.

7. Minimum Temperature

The minimum Temperature of 1% at the beginning of the Morning resulted in paying more interest than necessary at lower Maximum Temperatures.

8. Maximum Temperature Increases

The Maximum Temperature increased too quickly below peg, which disincentivized lending to the protocol in an efficient fashion and resulted in the protocol overpaying for loans.

9. Excessively Low Pod Rate Threshold

The Excessively Low Pod Rate threshold was too high.

Solutions

1. Rain Stalk Preservation

• Update LibSilo.transferStalk such that rainRoots are transferred rather than burned upon a Deposit transfer.

• Update LibSilo.burnActiveStalk to return deltaRainRoots, which corresponds to the number of Rain Stalk removed when burning active Stalk.

• Add a new LibSilo.mintRainRoots function that restores Rain Stalk to the user that performs the Convert.

• Update LibConvert._withdrawTokens to return the Rain Stalk removed from the user.

• Update LibConvert._depositTokensForConvert to take in deltaRainRoots, and to call LibSilo.mintRainRoots, if deltaRainRoots > 0 and the output is non-Germinating.

2. AL2L Converts

Re-add ANTI_LAMBDA_LAMBDA as a valid Convert type with a fix to the bug that removed all of the Rain Stalk during an AL2L Convert instead of reducing it by the appropriate number.

3. gm Reward

Update LibIncentive.fracExp to properly return the scaled reward when the gm function is 239 to 240 seconds late.

4. Flood Pods

• Add a new storage variable in the Rain struct, floodHarvestablePods.

• Update LibFlood.handleRain to reset floodHarvestablePods to 0 each Season, and LibFlood.floodPodline to store the actual Pods made Harvestable by the Flood.

• Update Sun.stepSun to always use floodHarvestablePods when calculating the newly Harvestable Pods when issuing Soil above peg.

5. Soil Issuance Above Peg

Update soilCoefficientLow in the evaluationParameters struct from 1.5e18 to 1.2e18 (i.e., 1.5x -> 1.2x). Update Sun.setSoilAbovePeg to apply the 0.5x Soil multiplier only when the Pod Rate is Excessively High. When the Pod Rate is Reasonably High and Pinto’s price is above the excessive-price threshold, issue the unadjusted Soil quantity instead of halving it.

6. Morning Auction Duration

Update delta in LibDibbler.morningTemperature such that each interval during the Morning lasts 24 seconds (12 blocks) instead of 12 seconds (6 blocks), extending the length of the Morning from 5 minutes to 10 minutes.

7. Minimum Temperature

• Update LibDibbler._scaleTemperature such that there is no longer a Temperature floor (previously set to 1%).

• Update LibDibbler.morningTemperature such that when delta == 0, the function returns 1% of the Maximum Temperature.

8. Maximum Temperature Increases

• Replace T_PLUS_3_L_MINUS_FIFTY, T_PLUS_3_L_PLUS_ONE and T_PLUS_3_L_PLUS_TWO in LibCases with T_PLUS_2_L_MINUS_FIFTY, T_PLUS_2_L_PLUS_ONE and T_PLUS_2_L_PLUS_TWO, respectively, which decreases the absolute Maximum Temperature increases from +3 percentage points to +2 percentage points.

• Update s.sys.casesV2 with the new case values.

9. Excessively Low Pod Rate Threshold

Update podRateLowerBound in the evaluationParameters struct from 0.05e18 (5%) to 0.03e18 (3%).

Rationale

1. Rain Stalk Preservation

Preserving Rain Roots through Deposit transfers and Converts into non-Germinating Deposits preserves ownership of Flood proceeds while allowing those operations.

2. AL2L Converts

Restoring AL2L Converts with the Rain Stalk fix restores their functionality without removing all of a user’s Rain Stalk.

3. gm Reward

Correcting the reward calculation preserves the intended incentive to call gm throughout the 239-to-240-second interval.

4. Flood Pods

Including Pods made Harvestable by a Flood keeps above-peg Soil issuance consistent with the total number of Pods made Harvestable.

5. Soil Issuance Above Peg

There is no need for the protocol to so aggressively try to increase the Pod Rate when the protocol has an excessively low Pod Rate. Removing the unintended 0.5x multiplier approximately doubles Soil issuance in the affected Reasonably High Pod Rate cases, restoring the quantity calculated from newly Harvestable Pods and the Maximum Temperature. The 0.5x reduction remains in place when the Pod Rate is Excessively High, when the protocol more strongly limits new debt.

6. Morning Auction Duration

By doubling the duration of the Morning Auction – and each interval of the auction – farmers will have more opportunity to participate manually, which should generally result in the protocol having to pay lower interest rates. A longer Morning Auction has minimal downside because the protocol will still offer Soil at the Maximum Temperature for 50 minutes during each Season, which leaves ample time for farmers to participate once the Morning Auction ends.

7. Minimum Temperature

The hard set minimum of 1% Temperature during the Morning Auction reduced the effectiveness of the auction during periods where the Maximum Temperature was very low. In instances where the protocol has little to no outstanding debt, it is highly plausible that some farmers would be willing to lend to the protocol at less than 1%. This minimum made it such that even if farmers were willing to do so, they would be unable to, resulting in the protocol potentially overpaying to issue debt.

8. Maximum Temperature Increases

The protocol raised the Maximum Temperature too quickly. By changing the Maximum Temperature increases from 3 percentage points to 2 percentage points it should decrease the rate the protocol needs to pay to attract creditors without decreasing its ability to do so.

The protocol was experiencing a nearly constantly increasing Maximum Temperature, which poses a short term threat to the protocol because it creates an incentive for potential creditors to wait for a higher interest rate, and a long term threat because if the Maximum Temperature increases in perpetuity, the protocol may eventually collapse under the weight of excessive outstanding debt.

By lowering the rate at which the Maximum Temperature increases, it should create a more efficient market for Soil in the short term and give contributors more time to design a better mechanism to prevent runaway Maximum Temperature in the long run.

9. Excessively Low Pod Rate Threshold

The Excessively Low Pod Rate threshold indicates the rate below which the protocol strongly prefers to take on more debt. It is likely safer for the protocol to be slightly more conservative and set this threshold to be 3% instead of 5%.

Contract Changes

Initialization

Executed InitPI3.init() at 0xa35c864c9154ef9fccdc037c489ee8bc30631477. It updated podRateLowerBound from 0.05e18 (5%) to 0.03e18 (3%), soilCoefficientLow from 1.5e18 (1.5x) to 1.2e18 (1.2x), and s.sys.casesV2 with the revised Temperature adjustments. If lastSopSeason equaled the current Season, it added totalSupply / 1000 to Soil.

Facet Changes

ApprovalFacet, ConvertGettersFacet, ClaimFacet, SiloFacet, SiloGettersFacet and SeasonGettersFacet were replaced without changes to their facet source files. Shared-library changes are described below.

Other Contract / Parameter Changes

Externally Linked Libraries

The table lists the external library deployments linked by the replaced facets. Where previous deployments differed by facet, the previous address is labeled with its consuming facets.

FieldFacet, ApprovalFacet, ConvertGettersFacet and SiloGettersFacet had no externally linked libraries in either implementation.

LibPipelineConvert, LibTokenSilo, LibEvaluate, LibGauge, LibShipping, LibWellMinting and LibGerminate were redeployed with unchanged library source files. Their new addresses do not indicate additional source changes beyond those described below.

• Doubled each Morning interval from 6 to 12 Base blocks (12 to 24 seconds), extending the auction from 5 to 10 minutes.

• Changed the initial Morning Temperature from a fixed 1% to 1% of the Maximum Temperature.

• Reduced selected Maximum Temperature increases from +3 to +2 percentage points. Initialization applied the 1.5x-to-1.2x low-Pod-Rate Soil coefficient and 5%-to-3% lower Pod Rate threshold described above.

Function Changes

• Updated LibSilo.transferStalk to transfer Rain Roots exceeding the sender’s remaining Roots to the recipient instead of burning them from the system.

• Updated LibSilo.burnActiveStalk to return deltaRainRoots in addition to roots. Updated LibConvert._withdrawTokens to return deltaRainRoots in addition to grown Stalk and PDV. Updated ConvertFacet and PipelineConvertFacet to pass this value through to _depositTokensForConvert, which calls the new internal LibSilo.mintRainRoots helper to restore burned Rain Roots for non-Germinating output.

• Restored ANTI_LAMBDA_LAMBDA dispatch in LibConvert.convert to LibLambdaConvert.antiConvert.

• Corrected the LibIncentive.fracExp branch cutoff from 240 to 238 seconds so the gm reward calculation covers 239–240 seconds.

• Updated LibFlood.handleRain to reset floodHarvestablePods to 0 each Season and LibFlood.floodPodline to store the actual number of Pods made Harvestable after the cap. Updated Sun.stepSun to include that count in above-peg Soil issuance.

• Added FieldFacet.floodHarvestablePods(), which returns the stored count.

• Updated Sun.setSoilAbovePeg to apply the 0.5x high-Pod-Rate coefficient when caseId.mod(36) >= 27 instead of >= 24. This excludes cases 24–26 within each 36-case group, where the Pod Rate is Reasonably High and Pinto’s price is above the excessive-price threshold. Those cases now use the unadjusted Soil quantity instead of half that quantity.

• Updated LibDibbler.morningTemperature to use 12 Base blocks per interval instead of 6. Removed the fixed 1% floor in LibDibbler._scaleTemperature. At delta == 0, morningTemperature returns 1% of the Maximum Temperature.

• Updated LibCases.setCasesV2 to use +2-percentage-point adjustments in the cases that previously used +3-percentage-point adjustments.

Event Changes

Existing event signatures were unchanged. The revised calculations changed the following emitted values.

• Incentivization now reports the corrected gm reward in beans for calls 239–240 seconds late.

• Soil reports soil issuance that includes Pods made Harvestable by Floods and applies the revised low- and high-Pod-Rate multiplier rules.

• Sow reports pods calculated using the revised Morning Temperature schedule and minimum.

• TemperatureChange reports +2 instead of +3 in absChange for the revised cases.

• TemperatureChange and BeanToMaxLpGpPerBdvRatioChange report caseId using the 3% lower Pod Rate threshold instead of 5%. The ratio-change event remains conditional on no oracle failure.