roles
roles copied to clipboard
bisq-network
Mediator
Docs: https://bisq.wiki/Mediator Team: @bisq-network/mediators
Proposal for @leo816's bonded mediator role: GitHub Issue Approved DAO Proposal Transaction
Proposal for @Bisq-knight's bonded mediator role: GitHub Issue Approved DAO Proposal Transaction
Proposal for @pazza83's bonded mediator role: GitHub Issue Approved DAO Proposal Transaction
I close it for now as it is not used atm. Can be re-opened once it becomes active.
Re-opening this issue, as it's the one we need to use, since it's in the code.
https://github.com/bisq-network/bisq/blob/master/core/src/main/java/bisq/core/dao/state/model/governance/BondedRoleType.java#L67
Cycle 9: I have mediated 65 cases and I have 9 cases pending in this cycle. I will make a breakdown of the cases in the coming days.
Cycle 8: I am a mediator and have a bond of 10 000 BSQ locked up. I have so far mediated 104 cases and have 8 cases pending. The cases started October 29. I don't know how I can prove this but I will discuss with the developers if a way can be found. For now it will be necessary to trust me. If I cheat I risk that my bond being confiscated. I will publish a report in a few days with a link here including a summary of the disputes.
In 16% of the cases the transactions was not published, but this problem has essentially disappeared with version 1.2.4 In 42% of the cases the seller released the BTC without further ado, often after a late payment. In 6% of the cases the buyer did not pay and stopped responding. In 5% of the cases the seller stopped responding and did not release the BTC despite being paid. In 8% of the cases the banks caused trouble, in a few cases due to "suspicious activity". In 2% of the cases the confirmation button of the seller did not work. These cases occurred in v1.2.3
The non-responding traders will likely not sign the mediators suggestion and are thus expensive for the DAO. If (when) the volatility goes up there will be many more forward trades and non-responsive buyers. If there was a way for disputes where the seller releases the BTC to simply go away it would help the mediators. I have a feeling that the number of disputes are decreasing as the traders move to v1.2.4.
I just looked up quickly the rough number of trades in that period and it was about 3000 trades. So if the other mediator has similar numbers we have about 210 disputes at 3000 trades which is 7%. This seems to be a bit above the average what I remember from the past (5%) but as we had several issues over the past weeks and as with the new trade protocol some cases are caused from usability/confusion this seems to be in the expected range. We have to work hard to bring those number down so user experience gets improved and costs for the DAO decreased.
I have done 165 mediations so far.
this overview was made 20 days ago:
an overview would be: -failed tx (not published) 35% -seller not responding 20%(usually they eventually respond and end up releasing) -bugs (like not being able to click on payment received or stuck on the blockchain confirmation part even though the tx has confirmations) 15% -banking issues 10% -people not wanting to complete the trade for various reasons 5% -users want to get their account signed 5% -Buyer used a different bank account to make the payment 10%
*the message issues have been completely solved, I believe all my messages have gone through and haven't been aware of lost messages from other parties.
*the invalid tx not getting published problem is also much better now but its still the biggest problem.
Since I made this overview 20 days ago, invalid transactions have significantly been reduced on my end, and I would say the biggest issue is when the BTC seller can't see the confirm payment received Button.
most recent overview would be:
failed tx (not published) 5% Seller not being able to see the confirm payment button 20% banking issues 10% both parties not wanting to complete the trade for different reasons 10% Buyer used a different name from the one shown on their Bisq account. 15% seller stopped responding and didn't release the btc, almost always ended up making a later payment 40%
It should be clear to any observer that the BSQ price and market has been very weak this month so far. Of course it's difficult to balance the supply and demand, and only 6 months ago I could barely find any BSQ to buy for my personal trading. But issuance has soared since then relative to the amount being burned in fees at least. So it does makes sense to me to look closely at the operating costs that we are seeing in compensation requests. Maybe there are places to cut costs.
Not that I can say I understand the work required to mediate a case, except from the point of view of a customer in a mediation, but it struck me that 50 BSQ per case was quite a high rate to charge. Now we are seeing the monthly bills hitting in comp. requests and I can give some minor analysis.
Hypothetical User X (prefers to remain anonymous) made 225 trades and had 9 cases that went to mediation (or legacy arbitration), none due to negligence, though 1 or 2 to software errors the prevented the trade from completing.
A mediation rate of 4% and a cost to BISQ of 450 BSQ. Fees paid in BSQ and revenue to BISQ were 3,050,000 sats all on the maker side. So assume 3x for the taker side, 9,150,000, arriving at 12,200,000 sats total. At 8500 sats per BSQ this is 1,435 BSQ.
So in an average case, the cost of mediation alone is already a huge chunk of fee revenue. And because it's a variable cost, it will scale more or less with exchange volume growth. So I think it's quite important to carefully consider the cost of that role.
Mediations are less and less every cycle, with our feedback programers are dealing with the different bugs that are appearing. I believe our role to be critical but also fading. The end goal is to have no mediation. @ifarnung rnung I believe your analysis not to be precise to the fact that you are simply putting out a single case from a trader and using it as the norm. The vast majority of users never have any issues, and the ones that do are every month less and also helped by us the mediators through keybase, please keep in mind that this rate also considers this kind of support and is not asked for anywhere else.
@ifarnung Thanks a lot for your analysis! You arre absolutely right we need to get mediation (and even more refund agent) cases down. They are a major cost factor (also with legacy arbitration). To fix open bugs, improve UX and finetune parameters like security deposit to bring that down is essential. Once bugs are fixed we should also start to extend the mediators group but onboarding mediators is still quite a bit of an effort and I think we should rather invest that effort in improvements to reduce cases before extending mediators.
To help to get there we need more detailed reports and better communication from mediators/refund agent to developers.
Cycle 11 I have mediated 313 cases since I started as a Mediator, On the previous cycle the biggest issue we faced was the bug that had to do with the seller not being able to confirm that he had received the payment, even though the buyer had clicked on "payment made" correctly. On this cycle the main concern relies on the seller not confirming the mediation ticket proposal. Knowing they will lose the deposit, they still disappear after they've received the fiat/altcoin payment from the btc buyer. This does not have an obvious solution but we are currently working on the support team to solve this as soon as possible.
an overview of the reasons why the tickets were opened in the first place would be:
-Failed tx: 20% -Seller makes a late confirmation: 20% -Seller never answers: 10% -Banking issues: 25% -User's not having the correct information or using the wrong bank account to make the payment:10% -seller not being able to confirm the payment was made: 10% -Both parties agree to cancel because a mistake was made: 5%
Notice the issue regarding the Sellers not being able to confirm still stands (although significantly reduced) but that is because there have been some cases where users that have opened the dispute had to wait 24hrs for the refresh button to appear so decided to open an early dispute and solve the problem faster. that waiting time will be cut down on future updates.
Invalid txs have been considerable this cycle although the number has not increased dramatically as you might think from comparing from last evaluations, I believe on the previous assessment I was only referring to the later part of the cycle and hence wasn't an accurate representation.
313 cases this cycle is a HUGE amount of cases and exactly the worry I was highlighting a few months ago about the very high costs of Mediation based on this roles compensation of 50 BSQ per case. I'm sure that you are providing excellent resolution of all these errors and failed trades, but can the DAO afford to pay 15,000+ BSQ this cycle for your support efforts? It just doesn't seem proportional at all to the amount of work done. In my country, 50 BSQ is almost a day's wage for white collar work, so we must admit this isn't a year's worth of work in one cycle... I understand that the first mediator set the bar at 50 per case, and you were just following, and I am sure some discussions have been had on how to budget this expense with the new Team Lead system. Anyway, often silence on BISQ means consent once it comes time to vote so I will voice my objection early so that at least people can think over how it should be handled going forward.
313 cases this cycle is a HUGE amount of cases
This is a misunderstanding; 313 is the total number of cases @leo816 has mediated since he has been a mediator. From his report for cycle 11 at https://github.com/bisq-network/compensation/issues/495#issue-564822991:
I have mediated 313 finished cases so far. [...] 61 on February 22 on March so far
(@leo816, it would be better to do everything in terms of cycles, not months, but that's a nit)
50 BSQ [per case is too much ...]
Also as mentioned in @leo816's cycle report, he is requesting 4100 BSQ based on the 50 BSQ per case figure, because this was the agreed upon figure when we decided to split up his compensation request in cycle 10 over two cycles. Going forward, he will request compensation at the rate of 25 BSQ per case:
As discussed with @cbeams this is the last cycle using 50 BSQ/ case [...] From the next cycle onwards, the price used for compensation per case will be cut in half to 25 BSQ.
And actually, I want to make a correction to that understanding Leo and I established. As discussed in @ifarnung's own proposal to become a mediator at bisq-network/proposals#184, we should not be denominating the value of work performed in BSQ. It should always be denominated in USD and then converted to BSQ at the published 90-day moving average for the current cycle. So, leo, if we do that rough calculation, you've been receiving 50 BSQ / case during a period where the 90-day average has been around 0.66 USD/BSQ. That equals 33 USD per case. Adjusting this down by half 50/2 = 25 BSQ per case, we get 33 USD/2 = 12.5 USD per case. @ifarnung is requesting 1000 USD/cycle in his proposal to become a mediator, and that would be the equivalent of 80 cases, which is higher than the current average number of cases per mediator (with 2 active mediators). From a budget perspective, I am fine with the agreed-upon 12.5 USD / case, and I would also be fine with a fixed amount per month as @ifarnung is suggesting. Assuming that @ifarnung's proposal gets approved and we go with 1,000 per month there, I'd let you choose, @leo816, whether you'd like to agree to the same. There is a lot of value in having competent mediators at the ready regardless of how many cases are coming in, and I believe that paying a steady rate for that is probably a better approach than paying per-case.
In any case, these changes were made in part because of your feedback, @ifarnung, so thank you for that. Note that the other mediator, @bisquser, is revoking their key, and so there won't be any contention about the 50 BSQ / case figure there. So we're basically dealing with a fresh start now.
I've updated the description of this issue to link to the new https://bisq.wiki/Mediator role documentation that @leo816 has been putting together in bisq-network/admin#26.
Sorry for my several misunderstandings, yes the 313 number being many months of work makes much more sense. Just to be clear, I am not asking to become a mediator, I'm far from skilled or capable from the providing that service. So once again, my thanks to @leo816 and @bisquser for the excellent and skilled work they performed in that regard. My critique was only the overall budget expenditure for a scaling and ongoing cost. Agreeing to a 50% cut in pay is certainly a show of good will, helpful participation in the austerity measures, and is very much appreciated by the community I am certain. So thanks to all for working that out. And at the lower rate per case, and with a lower estimate of cases handled per cycle, I think it does make more sense to compensate as you guys have agreed. Thanks for the update, another confidence boosting message to stakeholders. :)
Just to be clear, I am not asking to become a mediator
Ugh, my bad. I thought that was your proposal and see now that it is @wiz's. In any case, my comments based on it still stand, but thanks for clarifying, @ifarnung.
we get 33 USD/2 = 12.5 USD per case.
should be 16.5 no? which means that 60 cases would be around that 1000$ mark, sounds about right. I have no problem with changing to a steady rate but my question now would be, how do I value all the 149/313 mediation cases (+ all the cases in between now and the next cycle) that I haven't gotten compensation for yet? Im guessing that it would fall under the 25 BSQ/16.5 USD agreement we made. I think is best if I continue with the 16.5 USD until I catch up at least, but im open to discussion.
Hah, that was a terrible calculation on my part, thanks for catching it.
On Mar 12, 2020, at 7:12 PM, leo816 [email protected] wrote:
we get 33 USD/2 = 12.5 USD per case.
should be 16.5 no? which means that 60 cases would be around that 1000$ mark, sounds about right. I have no problem with changing to a steady rate but my question now would be, how do I value all the 149/313 mediation cases (+ all the cases in between now and the next cycle) that I haven't gotten compensation for yet? Im guessing that it would fall under the 25 BSQ/16.5 USD agreement we made. I think is best if I continue with the 16.5 USD until I catch up at least, but im open to discussion.
— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub, or unsubscribe.
Cycle 10-11: I have mediated 206 cases and I have 12 cases pending in the two last cycles. I have now revoked my role and will be replaced by different mediators, if the DAO so choses.
Cycle 11-12: 142 cases were mediated on March, that is double the monthly average due to the fact that the other mediator stepped down mid-cycle I was left with double the cases.
-this is an overview of the monthly cases up until now:
16 cases from the old mediation. 83 cases in November 62 on December 69 on January 61 on February 142 on March 38 on April so far
On the last cycle I mentioned that the main concern relied on the seller not confirming the mediation ticket proposal, knowing (or perhaps not fully knowing in most cases) they would lose their deposit, even then, they would still disappear after they received the fiat/altcoin payment from the btc buyer. This issue is still happening on this cycle and it does not have an obvious solution but we are currently working with the entire support team to solve this as soon as possible, specially given that we faced an attack and it passed as undetected for a week as a result of it being a recurring issue within the daily mediations.
Similarly, the main problem when talking about user-experience is that the seller can even confirm the payment after 9 days and would still get his funds without any penalty. That can't be the norm for a trader who is using the app correctly and is still being penalized. There has also been an increase in cancelled trades due to banks cancelling payments, specially with belle payments, there have been a lot of issues.
The new fixed rate of 1000 USD for the mediation role is done to ensure the mediator's interests are aligned with Bisq's. Before the 11th Cycle the price used for compensation per case was cut in half to 16.5 USD . Out of the 313 cases I had closed until then, only 162 were accounted for, corresponding to 2 cycles respectively, that leaves 151 cases unaccounted for or the equivalent of 2 cycles worth of cases. In the per-case scenario, that would amount to 2491,5 USD. Even though that would be the right price agreed upon for that time period, I believe that given the recent events on Bisq, its best to account it as just two cycles with the new fixed system, so that would equate to 2000 USD. The compensation for this cycle would then would be:
- 2000 USD from 2 earlier cycles to get up to date
- 1000 USD for cycle 12
Cycle 12 - Mediation Report
I'm still getting my head around on what is most relevant to report here.
Overview:
- 22 Cases in this cycle
- 15 closed
- 7 still open >>> either waiting on traders to respond or on 1.3.2 to be released (1 open is one of the victims of 7th of April)
Notable pattern: most cases actually are a result of fat fingers. Traders delay their payments or call mediation prematurely. The same people usually end up solving their own problems. I'm gonna gather some more data on that for the next reports.
Cycle 12 report
Handled 94 mediator tickets for 47 trades, only a few currently open pending nonresponsive trade partners waiting to begin arbitration
https://github.com/bisq-network/compensation/issues/547
Cycle 13 report [WIP]
Summary:
Total Cases in Cycle: 23
Still open: 4
Closed: 19
Total amount mediated: 2.75 BTC in trade amount (excluding deposits) Avg. case size: 0.12 BTC (average affected by a couple of large cases, most cases are 0.01 BTC) Avg cost per case: 43.88 USD
Overview (facts):
- Buyer delay is still the most common cause for cases
- the removal of mediation from the trade window as the first "get support" behavior did reduce the fat finger cases
Avg. Buyer account age (for fiat trades, and trades up to 0.01 BTC) (21 trades, 58% of all cases)
| Buyer median age | Seller median age | |
|---|---|---|
| Cycle 12 | 17 | 38 |
| Cycle 13 | 22 | 62 |
The same view for trades above 0.01 (15 trades)
| Buyer median age | Seller median age | |
|---|---|---|
| Cycle 12 | 99 | 402 |
| Cycle 13 | 114.5 | 133.5 |
Opinion:
- the initial idea of having mediators paid a fixed amount is good for when there are many cases, but right now it is proving just marginally better for the DAO in terms of cost. I would suggest that, to reduce the operating costs, we change the payment structure for mediators:
20 USD per case or 1000 USD per month, whichever is lower. 20 because it is significantly cheaper for the DAO (at least from my standpoint) and still compensates mediators for the time spent troubleshooting users yet keeps capacity available for when volume demands it without having the lead time to propose a role, do the bonding and onboard/train mediators when needed. - Most cases are caused by new users (most likely buyers). If we increase efforts in education we may be able to reduce mediation requests. (see tables above)
- to reinforce the point above, many trades get completed without the mediator "mediating", but only explaining/educating users. So if we were to add the Wizard or some more content to the app/social media where users could help themselves we could potentially reduce cases even further, reducing costs even further.
@Bisq-knight, is the 402 value for seller median age a typo?
Nope, is actually the value that divides the larger fiat trades in the last cycle: This is the array of ages for this cohort: [18, 45, 226, 402, 527, 532, 789]
Cycle 13 report
Total Cases in Cycle: 48
Still open: 9
Closed: 39
- Most cases are caused by new users (most likely buyers). If we increase efforts in education we may be able to reduce mediation requests. (see tables above)
I agree completely with this point but I believe the solution is to have a "most asked questions" bullet point list filled with the most common issues new users face and have that as a way to avoid some of the pointless mediations we are seeing.
Steve Jain suggested that I change my language to Spanish, Once I change my data directory I will also add Spanish as a language, In case I can only chose one language then might be interesting to have one of the mediators who can speak this language, giving that option to users who might find it easier to communicate in their native language.
-
On my case the most common issue I see is still one of the parties not responding, usually the seller.
-
A big issue we still haven't taken measures over is the fact that a buyer is able to wait 10 days to make a payment and he will receive no penalty, similarly a seller can wait many days before releasing the btc an as long as the funds don't get to the donation address he will receive his funds like nothing happened. Whenever this happens this makes it a huge impact on the user who is obviously frustrated and has been penalized with time when they did nothing wrong.
Cycle 13 report
- Bisq loses some mediator ticket messages and Keybase must be used often
- Most mediation cases are useless as mediator cannot do anything
- Most disputes caused by unresponsive trade counterparties
- Users always solve themselves or one party proceeds to Refund Agent arbitration
- This role is a bit frustrating as users are not happy with Bisq experience currently
/cc https://github.com/bisq-network/compensation/issues/569
Cycle 14 report
Summary:
Total Cases in Cycle: 30
Still open: 7
Closed: 23
- definition: I made the previous report but didn't clarify on my definition of what's in a given cycle. I'm using the mediation start date as the main date AND while the voting period is open, the case falls into that cycle. Because we need to do these reports in the middle of the voting period I think it is worth to just update the reports once the next report is done.
With that in mind here is the consolidated view on Cycle 13 and the WIP for Cycle 14:
Cycle 13 [Consolidated]
Total Cases in Cycle: 36
Amount mediated: 3.4068 BTC
Notes: what I wrote before
Cycle 14 [WIP]
Total Cases in Cycle: 30
Amount mediated: 2.7203 BTC
Notes: we had quite a few +6 cases related to Revolut. This was mainly due to Revolut updating their design and breaking the flow that bisq traders would normally do. Now the solution is very simple: ask the trader for the username and search with that, also, keep in mind that usernames ARE case sensitive
Cycle 14 Report
Total Cases in this Cycle: 38
Still open: 10
Closed: 28
Because we need to do these reports in the middle of the voting period I think it is worth to just update the reports once the next report is done.
I agree with this proposal and I will analyze the cycle up until today, will update it at the end of such cycle with the remaining information.
-
35% Banking issues (Revolut, Zelle, SEPA)
-
15% both parties agreed to cancel
-
35% BTC seller took longer to answer/make the payment
-
15% Unexperience, parties unsure of how the process works
This cycle we've had a considerable amount of issues with Revolut between @Bisq-knight and me, the problem relies on revolut not recognizing some of the phone numbers as having accounts on the platform. Usually the accounts are fairly new. The solution we've suggesting is creating a payment link/ using the username (@example) or if none of these work just making a normal transfer using the IBAN revolution provides.
Another pending issue is the problem with users who never respond and trades end up going to arbitration which makes no sense for the other party that often has done everything correctly. Similarly a BTC seller may release the funds after 10 days with no penalty at all, there must be some way to control this.
Cycle 14 report
- Upgraded to a Tor V3 onion for my mediator node to resolve mediator mailbox message issues
- Handled 26 mediation tickets in June so far, seems down from last month
- Handled ~100 mediation tickets in May
- Most cases are same as leo's report above: Revolut, bank name not matching, or future trades
/cc https://github.com/bisq-network/compensation/issues/602
Cycle 15 Report
Total Cases in this Cycle: 23
Still open: 6
Closed: 17
New action protocol was proposed.
I will analyze the cycle up until today, will update it at the end of such cycle with the remaining information.
- 15% both parties agreed to cancel
- 20% Trader provide the wrong information
- 40% BTC seller took longer to answer/make the payment
- 5% Unexperience, parties unsure of how the process works
This cycle I've had nearly no banking issues in comparison with cycle 14. Specially with Zelle and Revolut where we've had most problems lately. I've had an issue with a user that could not click on "payment started" button. Users have seemed to have learnt to solve the Revolut issue by creating a payment link/ using the username (@example) or just making a normal transfer using the IBAN revolution provides.
A clear take away from this is that most mediation tickets arise from users being impatient, they seem unnecessary and we will keep a close look into the following cycles to see if this trend continues.
Another pending issue is the problem with users who never respond and trades end up going to arbitration which makes no sense for the other party that often has done everything correctly. Similarly, a BTC seller may release the funds after 10 days with no penalty at all, there must be some way to control this. (still working on this)
Cycle 15 Report
Total Cases (so far):
Open cases: 4 (2 from cycle 14 and 2 from cycle 15)
Closed: 20
Amount mediated: 0.9389 BTC
I'll write up the summary and analysis once the cycle is closed