isabella232 / retroactive-query-1

@pooltogether/retroactive-query

This repository contains queries that produce the tables of retroactive POOL token distributions.

The queries run in Google BigQuery against the bigquery-public-data.crypto_ethereum dataset.

Data for this dataset is extracted to Google BigQuery using blockchain-etl/ethereum-etl.

Specifications

There are 3 versions of the PoolTogether protocol. The cutoff blocks are as follows:

• v1: fixed period of 43,200 blocks (1 week - v1 pools were ephemeral)
• v2 and v3: block number 11,656,283

Total distribution is aimed at 1,500,000 POOL.

All users

20 POOL goes to:

• any account that directly withdrew or deposited into any version of the protocol, including the v2 pods

Liquidity providers

• all liquidity is weighted by the time (measured in blocks) multiplied by the amount expressed as a percentage of total liquidity to give a total_lp_shares_fraction per address
• The additional rewarded tokens is then calculated as: LOG2(1 + total_lp_shares_fraction * 10000)/lp_share_total) * remaining_lp_tokens where remaining_lp_tokens is the number of tokens after the base reward of 20 POOL for each address has been subtracted from the total distribution.
• This log influenced distribution allows for a distribution that is reduced for large providers and enhanced for smaller holders

Reproduction

You can reproduce the results of this query by forking this repository and adding your own secrets to run in your own GCP account.

1. Create a Google Cloud project here
2. Find your Project ID in the Google Cloud console here
3. Create a Service Account in GCP and download the credentials JSON file. See here
4. Fork this repository on Github
5. Add the Github project secret GCP_PROJECT_ID. The value should be your GCP Project ID from earlier.
6. Add the Github project secret GCP_SA_KEY. The value should be the content of the GCP Service Account JSON credential file.
7. Go to the actions tab of your fork
8. Run the workflow (roughly ~12 minutes to complete)
9. Inspect the resulting tables - all_earnings_hexadecimal is the final table used in the merkle distribution.
10. Create a query that selects all rows from the all_earnings_hexadecimal table and orders by address ascending.
11. Export the query result as a JSON file.
12. You can now verify that the file matches the merkle distribution in the [project] (https://github.com/pooltogether/merkle-distributor)

Determinism notes

Note that, for floating point input types, the return result of aggregations is non-deterministic, which means you will not get the exact same result each time you aggregate floating point columns.

These queries make use of floating point numbers. However in the final all_earnings_hexadecimal table, we truncate to 6 decimal places so that the result used for production is the same across multiple runs.

See https://cloud.google.com/bigquery/docs/reference/standard-sql/aggregate_functions for more information.