Replies: 5 comments 13 replies
|
On mainnet the parameters are different and you would definitely not get the same results fee-wise. Thank you for the detailed feedback, that is very much appreciated. @zkSNACKs/visual-design-group @zkSNACKs/code-team |
|
For your info, in addition to the above: I have left the privacy boost active, and it coinjoined all funds further down to the point where there are now only 2 utxo's of 5.000 sats and one of 4.718. Meaning, it 'cleaned' away 75% of the funds! Would this also be different in mainnet? What is the reason of the difference in parameters for testnet and how could be better test the actual behaviour? |
|
Hi, testing the testnet release now, and I still find the same costly behaviour: I have a set of 4 inputs: 1000sats, 1100sats, 5000sats and 6700sats. There were already 3 rounds of coinjoins with a total fee of 4800sats, which is around 35% of the inputs. And it is still trying to coinjoin as privacy is around 80% 2 other inputs (100000 + 10000 sats) were fuly coinjoined just before in 1 round for only 1700 sats, which is very reasonable. This is not very consistent and I believe especially for very small inputs very costly. If it is not feasible to coinjoin these under a certain % cost maybe they should be flagged and excluded instead of 'burned'? |
|
One additional thing: you can choose to only coinjoin when fees are cheeper than the median of the last day, week, or month. This means you only participate in coinjoins when the fees are relatively low, and you don't participate in fee spikes. In general, if you are concerned about fees, then select the |
|
This discussion is now locked, as feature/enhancement requests are being moved to the Issues tab for further processing. Please use #9814 for future comments regarding this. |

Uh oh!
There was an error while loading. Please reload this page.
Hi,
While trying out wasabi 2.0 on testnet I noticed some unexpected behaviour after clicking boost privacy.
1st finding
I pressed boost a few successive times because the gui didn't really show that it was starting coinjoining. Also, even while boost remains activated, the spinning wheel and the 'coinjoinging' label in the menu disappears between the actual coinjoins, giving me the idea that it stopped boosting the privacy.
I have opened the wallet in electrum as well, so I can track each coinjoin. While they appear there, the transaction list in Wasabi remains static and only shows the tx fees from the first join. It's not clear something has happened.
When opening the details, the other transactions id's are added there though. (btw, the total mining fee there far exceeds what I paid and is maybe the sum for all users?)
=> proposal: better explain what is going on in the gui, more status updates
2nd finding:
The coinjoins seems quite costly. I started with an utxo of 58.780 sats.
Wasabi performed these successive coinjoins:
58.780 to 50.000 and 8.192 (cost 588 sats= 1%)
-> 50.000 to 39.366 and 10.000 (cost 634 sats= 1.3%)
--> [39.366 + 10.000] to [32.768 + 8.192 + 5000] (cost 3.406 sats = 6.9%)
-> 8.192 in 5.000 (cost 4.192 sats= 39%)
The total cost so far is 8.820 sats or 15%.
Currently, it still says privacy level 0,00%. The # of outputs was quite small each time (9 or 10), so I think it is accurate that the privacy of these coins was not increased much. However, the user in this case paid a relative big cost for not having added value.
I assume there is some dust threshold playing here backend side, but in that case maybe there is no point in trying to coinjoin utxo's all the way down till the dust limit (eg 8000 sats to 5000)?
=> proposal: avoid costly coinjoins (small ones with dust thresholds?); allow (user selectable) fee rate limits; only coinjoin when enough outputs so it's worth the mining fee; maybe think about simulation if technically feasible?
I do not know if the behavior is different on mainnet due to larger available volume and maybe less pronounced with bigger utxo's, but thought to share these findings here.
All reactions