dotnet run -- --rpcuser=user --rpcpassword=password
This repository is looking for answers on 2 questions:
- How much non-mixed bitcoins are entering Wasabi mixes?
- How much non-mixed bitcoins are leaving Wasabi mixes?
Question (1) provides a comparable metric to traditional centralized mixer volumes and a more accurate measurement of wallet adoption.
Question (2) aims to create a business case for better mixing algorithm, as the change outputs those aren't get re-enqueued to another mix are proof of impatient users and lost mixing opportunity.
2018-7 Fresh bitcoins: 1, Nonremixed change: 0 Nonremixed <2 anons: 0 Nonremixed <5 anons: 0 Nonremixed <10 anons: 0
2018-8 Fresh bitcoins: 16, Nonremixed change: 4 Nonremixed <2 anons: 4 Nonremixed <5 anons: 4 Nonremixed <10 anons: 5
2018-9 Fresh bitcoins: 49, Nonremixed change: 11 Nonremixed <2 anons: 11 Nonremixed <5 anons: 11 Nonremixed <10 anons: 14
2018-10 Fresh bitcoins: 67, Nonremixed change: 5 Nonremixed <2 anons: 5 Nonremixed <5 anons: 5 Nonremixed <10 anons: 5
2018-11 Fresh bitcoins: 3164, Nonremixed change: 2504 Nonremixed <2 anons: 2504 Nonremixed <5 anons: 2504 Nonremixed <10 anons: 2504
2018-12 Fresh bitcoins: 1272, Nonremixed change: 419 Nonremixed <2 anons: 419 Nonremixed <5 anons: 419 Nonremixed <10 anons: 419
2019-1 Fresh bitcoins: 2969, Nonremixed change: 428 Nonremixed <2 anons: 478 Nonremixed <5 anons: 563 Nonremixed <10 anons: 672
2019-2 Fresh bitcoins: 2235, Nonremixed change: 313 Nonremixed <2 anons: 381 Nonremixed <5 anons: 510 Nonremixed <10 anons: 609
2019-3 Fresh bitcoins: 3553, Nonremixed change: 221 Nonremixed <2 anons: 378 Nonremixed <5 anons: 526 Nonremixed <10 anons: 611
2019-4 Fresh bitcoins: 3435, Nonremixed change: 487 Nonremixed <2 anons: 558 Nonremixed <5 anons: 658 Nonremixed <10 anons: 778
2019-5 Fresh bitcoins: 2988, Nonremixed change: 278 Nonremixed <2 anons: 313 Nonremixed <5 anons: 412 Nonremixed <10 anons: 543
2019-6 Fresh bitcoins: 5288, Nonremixed change: 513 Nonremixed <2 anons: 661 Nonremixed <5 anons: 951 Nonremixed <10 anons: 1165
2019-7 Fresh bitcoins: 3754, Nonremixed change: 255 Nonremixed <2 anons: 282 Nonremixed <5 anons: 347 Nonremixed <10 anons: 436
2019-8 Fresh bitcoins: 20328, Nonremixed change: 1481 Nonremixed <2 anons: 2276 Nonremixed <5 anons: 4251 Nonremixed <10 anons: 10191
2019-9 Fresh bitcoins: 13456, Nonremixed change: 506 Nonremixed <2 anons: 857 Nonremixed <5 anons: 2109 Nonremixed <10 anons: 4632
2019-10 Fresh bitcoins: 4802, Nonremixed change: 215 Nonremixed <2 anons: 252 Nonremixed <5 anons: 340 Nonremixed <10 anons: 495
2019-11 Fresh bitcoins: 4295, Nonremixed change: 371 Nonremixed <2 anons: 424 Nonremixed <5 anons: 524 Nonremixed <10 anons: 618
2019-12 Fresh bitcoins: 5617, Nonremixed change: 504 Nonremixed <2 anons: 1152 Nonremixed <5 anons: 1537 Nonremixed <10 anons: 1781
2020-1 Fresh bitcoins: 8125, Nonremixed change: 969 Nonremixed <2 anons: 2486 Nonremixed <5 anons: 3182 Nonremixed <10 anons: 3501
2020-2 Fresh bitcoins: 7644, Nonremixed change: 1408 Nonremixed <2 anons: 2401 Nonremixed <5 anons: 3013 Nonremixed <10 anons: 3292
2020-3 Fresh bitcoins: 9489, Nonremixed change: 961 Nonremixed <2 anons: 2092 Nonremixed <5 anons: 2901 Nonremixed <10 anons: 3403
2020-4 Fresh bitcoins: 7651, Nonremixed change: 636 Nonremixed <2 anons: 1314 Nonremixed <5 anons: 1817 Nonremixed <10 anons: 2185
Fresh bitcoins coming into Wasabi every month. This is the most accurate measure of the adoption of the wallet I can think of.
A user who mixes does not always get back mixed coins. However these unmixed changes should be enqueued into another mix in order to be mixed. The graph illustrates how many unmixed bitcoins aren't enqueued to another mix. It also shows non-remixed coins - those only achieved 2, 5, and 10 anonymity sets. This is a lost opportunity for Wasabi. It is a proof of inefficient mixing algorithm.
This is a derived metric from fresh bitcoins coming into Wasabi and non-remixed changes leaving Wasabi. It shows the inefficientcy of Wasabi percentage.
For example in December: 9% of coins those participated in Wasabi mixes have never been mixed, 20% of the coins were only been mixed with 2 or less anonymity set, 27% of the coins were only been mixed with 5 or less anonymity set, 31% of the coins were only been mixed with 2 or less anonymity set.
Also note that, since remixing is impossible to detect, the larger the threashold grows, the more inaccurate these reults are. While I would be interested in how many of Wasabi coins were mixed with 50 anonset at least, I belive calculating it with my current algorithm would yield inaccurate results.