Trust Feature idea: give DT1 the ability to remove specific feedbacks from DT

[nakamura12]

I think if a DT1 have this ability then I think it can be abused to remove a feedback that shouldn't be deleted or removed or maybe use it to earn money (payment in crypto) with it like a forum member will pay crypto like Bitcoin to the DT1 to remove a feedback. Maybe It will happen or maybe not in my opinion.

Perhaps requiring maybe 5 DT members to agree with a down vote before confirming it would be a good idea.

5 DT’s agree & click down vote ensures potentially harsh or unfair neg is not shown by default.

If DT1 have the ability to remove feedbacks and having other DT1 vote for it before the feedback is removed in my opinion is helpful and avoid a DT1 member abuse the system. I would definitely agree with this. I was thinking if all DT1 member should vote for it except in situations where a DT1 is the one that sent the feedback because I am sure that DT will oppose the removal of the feedback that he/she sent to that forum member.

[1714252125]

1714252125

[1714252125]

1714252125

[1714252125]

1714252125

[1714252125]

1714252125

[jokers10]

The case in Reputation is a good example: topics about similar issues are popping up, again and again. And this won't be the last time. I don't remember how often the topic has popped up...
By voting out certain feedbacks, we might avoid these cases to pop up again and again because we have an initial decentralized DT decision and everyone has to deal with it. Going against DT's decision is just not profitable for most cases.

Well, when we talk about DT we mean that every DT member understand the issue, makes own conclusion and then acts. In theory. So if we'll look at the situation with flags we'll see that there are usually just few votes from DT, because others don't want to waste their time on investigating each case and some are voting just because they believe some other reputable DT members understood the issue enough and made a correct choice.

Spreading of this practice on votes implies that DT members will find additional time on researching many other cases to make own conclusions in each of them. Or will put together a group of like-minded and vote for each others' tags and against tags which are not pleasant for one of them. And we'll see much more topics in Reputation to vote for and against.

Basing on experience with flags I can say that only minority of DT will vote for and against others' tags. So if someone will get a negative tags and will ask for help some known compassionate DT members in PM, he'll get an opportunity of cleaning the negative tag he got and probably no one will even see this. Especially if a DT member who left that tag is on a vacation or left the forum.

And next. When we count votes, how will we count: if someone is in DT this month but not in DT next month, will a tag become trusted and untrusted like now basing on who is in DT? Reviews becoming more and more flickering... it could destroy Trust system reputation. As for now everyone in DT has his own reputation, and if everyone would have a right of voting for each tag, then there will be a common reputation. Each one will have an opportunity to say: yes, I've made a mistake, but what others did at that moment? I'm not fond with idea to bear some part of responsibility for cases I didn't investigate enough to vote or leave my own tag. I don't have time on deep investigation for each tag of everyone, each flag, etc. And to vote, I think the one should make own investigation of a case.

As I already said, I see we have some problems, but I suppose we'll get even more problems if to change a system this way.

[Pmalek]

Really?

I don't want to play devil's advocate here, because although it seemed like a good idea initially, and then seeing the arguments presented, I think it is better to leave things as they are, but when you said this I went to look at the first case that came to my mind, I guess because it was the most striking recently.

https://bitcointalk.org/index.php?action=trust;u=140827

There is not a single vote against the flag.

LoyceV understood what I was trying to say. I didn't mean an 80-90% majority from those that voted, but a majority from the whole list of DT1s. Meaning from the list of 100 DT1 members. And LoyceV is right when he says it would be borderline impossible to get such a majority vote here. Even in a clear case like yogg's rug pull.

Maybe the criteria should be that 2/3 or 3/4 of all DT1s must vote. And then if that criteria is met, there must be a majority vote of 80-90% in favor of removing the trust rating.

[dragonvslinux]

Really?

I don't want to play devil's advocate here, because although it seemed like a good idea initially, and then seeing the arguments presented, I think it is better to leave things as they are, but when you said this I went to look at the first case that came to my mind, I guess because it was the most striking recently.

https://bitcointalk.org/index.php?action=trust;u=140827

There is not a single vote against the flag.

LoyceV understood what I was trying to say. I didn't mean an 80-90% majority from those that voted, but a majority from the whole list of DT1s. Meaning from the list of 100 DT1 members. And LoyceV is right when he says it would be borderline impossible to get such a majority vote here. Even in a clear case like yogg's rug pull.

Maybe the criteria should be that 2/3 or 3/4 of all DT1s must vote. And then if that criteria is met, there must be a majority vote of 80-90% in favor of removing the trust rating.

Trying to follow the alternating proposals here but failing to understand the logic of 2/3 or 3/4 of DT1 to vote. For starters, it's very unlikely to get 61 to 69 DT1 members to vote on anything imo.

I don't see the issue with having only 50%+ agreement as the threshold for having feedback removed, as long as there is the option to upvote in order to counter downvotes. I also don't understand why this is suggested for DT1 only instead of all DT. Many of the most trusted DT members aren't even DT1 anymore anyway, as have wiped trust list or opted for exclusion, but are their feedback is still highly trusted.

To me it makes more sense to keep it simplistic like with the trust flags, with the addition of requiring at least 5 DT votes prior to any removal of default feedback to take affect. After all DT/DT1 is determined by majority inclusion (even 50%) that determines whether feedback is trusted by default or not, so to me it makes sense that any feedback votes from DT/DT1 are treated in the same manner as inclusions/exclusions.

Either that or increase the voting requirement to 10 DT votes (as opposed to 5 DT1 votes) if you want to avoid too many feedbacks being countered which could well end up being the case with such a low threshold. This would require a minimum of 6 DT down votes (based on majority agreement) in order for feedback to no longer be trusted, or otherwise the full 10 DT downvotes if there is no opposition.

Also having only DT1 able to upvote/downvote means that some feedback will "flip flop" on and off a lot more than if the votes are based on DT, due to how the DT lottery works, which wouldn't be ideal imo.

[Stalker22]

Here is an idea. What if we completely remove the negative trust (making it neutral) and use only flags for trust issues and trade risk?

Flags don't apply to most of the users who I gave negative feedback. Take this guy for example.

They do not now, but they could if theymos makes it so. The guy from your example is a trade risk.

[philipma1957]

look at my trust

https://bitcointalk.org/index.php?action=trust;u=64507

I was given a negative trust by jvanname

not a good reflection of the trust system.

https://bitcointalk.org/index.php?action=profile;u=1043276

if you read about him and see his posts and his distrusts it is exactly why the trust system needs help.

Maybe a review committee of egregious posts and trusts.

BTW is he banned? he posted on sept 4 so I think he is active.

[LoyceV]

look at my trust

https://bitcointalk.org/index.php?action=trust;u=64507

I was given a negative trust by jvanname

not a good reflection of the trust system.

He's not on DT, so his feedback isn't shown by default. At least this part of the Trust system works fine.
I guess it's your first negative, I remember those days, the first one hurts the most. After that, you stop caring. Look at The Sceptical Chymist's untrusted feedback for example. Just scroll down fast That spam is one of the reasons theymos made changes in the number of line breaks feedback can have.

[philipma1957]

look at my trust

https://bitcointalk.org/index.php?action=trust;u=64507

I was given a negative trust by jvanname

not a good reflection of the trust system.

He's not on DT, so his feedback isn't shown by default. At least this part of the Trust system works fine.
I guess it's your first negative, I remember those days, the first one hurts the most. After that, you stop caring. Look at The Sceptical Chymist's untrusted feedback for example. Just scroll down fast That spam is one of the reasons theymos made changes in the number of line breaks feedback can have.

I  have a few distrusts  buried deeply from when I used to try and scam bust bad sellers in the 2013-2015 timeline.

But that particular person jvanname is over the top abusive to everyone in every post.

If you look at his distrusts they are fairly ridiculous in every case.

More experienced members know that the distrust he gave is worthless but newer people do not.

So when you look at my trust feedback his is right on top 🔝 thus it is annoying.

Also any DT1 member can now include him at anytime and boom I have a fully visible distrust. Which would mean I would have to distrust the DT1 member that decided to trust him. And bla bla bla drama drama drama.

So when minor players do what he did it undermines the entire system.

Its a bit hidden because most Dt1 people are trustable.

please note most means 51% or more not 99%.

So to a large degree we need a mechanism to protect trusts from abuse.

Letting jvanname drop distrusts for poor reasons or no reasons weakens the system.

[LoyceV]

Also any DT1 member can now include him at anytime and boom I have a fully visible distrust. Which would mean I would have to distrust the DT1 member that decided to trust him. And bla bla bla drama drama drama.

Two users already distrust him, and if his feedback is inaccurate, you can exclude him for your Trust list already.
If DT1-members include the wrong users, they're supposed to be excluded. It's drama, but it comes with a decentralized system.

Letting jvanname drop distrusts for poor reasons or no reasons weakens the system.

He's not the only one, and I wouldn't say it weakens the system. It's just the result of not moderating feedback.

[1miau]

I want to comment on 1miau suggestion of 5 DT1s having the possibility to cancel out certain feedback by effectively moving it to the untrusted category.
I think 5 DT1s is a small number. I would support a majority vote in this case. I don't know how many DT1s we have, but I would want at least 80% if not 90% of DT1s voting in favor of removing a certain feedback.

Why? Because there shouldn't be a possibility that my 5 closest DT1 buddies can collude and remove feedback I don't want to see on my profile. But even if that happens for this DT1 cycle, what's going to happen with next month's change and reorganization of the DT1 list? Will there be new votes or is the decision made final?  

5 DT1 members was just a rough number, could be 10 or 15 as well.
Because most likely, not many DT members will read everything about the case and many DT1 members won't join the vote.
Yes, 5 DT1 members seems to be a low number but it's also not easy in my opinion to convince 5 DT1 members to vote in favor of removing a feedback. And if you have 2 DT1 members against it, you need 7 in favor already.

The case in Reputation is a good example: topics about similar issues are popping up, again and again. And this won't be the last time. I don't remember how often the topic has popped up...
By voting out certain feedbacks, we might avoid these cases to pop up again and again because we have an initial decentralized DT decision and everyone has to deal with it. Going against DT's decision is just not profitable for most cases.

Well, when we talk about DT we mean that every DT member understand the issue, makes own conclusion and then acts. In theory. So if we'll look at the situation with flags we'll see that there are usually just few votes from DT, because others don't want to waste their time on investigating each case and some are voting just because they believe some other reputable DT members understood the issue enough and made a correct choice.

That's why there was given a low number of 5 DT1 member votes. Many DT1 members won't bother to vote in it. But we can also increase it to 10 DT1 votes.

Spreading of this practice on votes implies that DT members will find additional time on researching many other cases to make own conclusions in each of them. Or will put together a group of like-minded and vote for each others' tags and against tags which are not pleasant for one of them. And we'll see much more topics in Reputation to vote for and against.

No need to require more DT1 members reading about it.
We can do it like flags, works pretty well currently.
Same procedure like flags and it's no issue at all.

Basing on experience with flags I can say that only minority of DT will vote for and against others' tags.

Yes and it's working pretty well for flags currently.  

And next. When we count votes, how will we count: if someone is in DT this month but not in DT next month, will a tag become trusted and untrusted like now basing on who is in DT?

There's no final decision on that but changes depending on who's DT1 currently are definitely a good idea because it's representing DT1.

Reviews becoming more and more flickering... it could destroy Trust system reputation.

Seriously, where should it "destroy" the trust system?
DT1 is currently determined that way, our own trust scores change by a small margin every week, depending on who's DT1 and DT2 and even DT2 is changing due to that already.
Nothing got destroyed from that.  

I don't have time on deep investigation for each tag of everyone, each flag, etc. And to vote, I think the one should make own investigation of a case.

No need for you to investigate such a case if you don't like to investigate. It'll be similar to flag support and it's working well currently.

[jokers10]

Basing on experience with flags I can say that only minority of DT will vote for and against others' tags.

Yes and it's working pretty well for flags currently.  

Flags have rather strict rules and feedback doesn't. Otherwise there was no need in flags, it could be done on a base of quantity of tags. Some tags are left on the very boundaries of a consensus of a Trust system; flags in the same cases would have been opposed, because of strict rules. If to make the same system as flags, then what will be the difference, why do we need two similar systems?

Reviews becoming more and more flickering... it could destroy Trust system reputation.

Seriously, where should it "destroy" the trust system?
DT1 is currently determined that way, our own trust scores change by a small margin every week, depending on who's DT1 and DT2 and even DT2 is changing due to that already.
Nothing got destroyed from that.  

I said "destroy Trust system reputation", not the system itself. It doesn't matter if you are in DT1 or in DT2, your tags will stay visible by default the same way. If the number of votes for and against tags will change each DT1 reshuffle, there will be flickering in feedback: some tags will appear while some will disappear. Does it look reasonable and reputable when the same tag of a same person is correct when there's one DT1 composition and becomes totally incorrect with another DT1 composition in a month and the same way each month?

I don't have time on deep investigation for each tag of everyone, each flag, etc. And to vote, I think the one should make own investigation of a case.

No need for you to investigate such a case if you don't like to investigate. It'll be similar to flag support and it's working well currently.

Right, no need for me, no need for someone else. So if someone will gather with several like-minded they can change a system as they will like, because majority won't get much additional time to do much additional work to monitor the correctness of Trust system work.

As for flag system: it is different. Tags are much less strict. I voted for 205 flags because it is easy in many cases, but I left just a 100 tags and big number of them is based on one big farm investigation.

[Stalker22]

Flags have rather strict rules and feedback doesn't. Otherwise there was no need in flags, it could be done on a base of quantity of tags. Some tags are left on the very boundaries of a consensus of a Trust system; flags in the same cases would have been opposed, because of strict rules. If to make the same system as flags, then what will be the difference, why do we need two similar systems?

But we already have different levels of flags, why not introduce just one more, for example, Flag Type 0 (which would replace neg. trust).

I said "destroy Trust system reputation", not the system itself. It doesn't matter if you are in DT1 or in DT2, your tags will stay visible by default the same way. If the number of votes for and against tags will change each DT1 reshuffle, there will be flickering in feedback: some tags will appear while some will disappear. Does it look reasonable and reputable when the same tag of a same person is correct when there's one DT1 composition and becomes totally incorrect with another DT1 composition in a month and the same way each month?

Again, we already have such a system with flags. Besides, I believe the number of such cases would be relatively low. Consider this scenario: if a flag receives the necessary support from DT1 members to become active, and one of those members later drops out of the DT pool, we have two options. We can either keep the flag active, or the flag becomes inactive until it secures the support of at least one more new DT1 member.

No need for you to investigate such a case if you don't like to investigate. It'll be similar to flag support and it's working well currently.

Right, no need for me, no need for someone else. So if someone will gather with several like-minded they can change a system as they will like, because majority won't get much additional time to do much additional work to monitor the correctness of Trust system work.

As for flag system: it is different. Tags are much less strict. I voted for 205 flags because it is easy in many cases, but I left just a 100 tags and big number of them is based on one big farm investigation.

But that is actually a good thing. Flags need thorough investigation and supporting evidence. That is why it is usually pretty straightforward to check out the evidence and investigation someone else has done and decide whether to support or oppose the flag. Tags, on the other hand, often get slapped on without much reference or clear evidence. Those kinds of tags should not be part of the trust system.

[philipma1957]

Also any DT1 member can now include him at anytime and boom I have a fully visible distrust. Which would mean I would have to distrust the DT1 member that decided to trust him. And bla bla bla drama drama drama.

Two users already distrust him, and if his feedback is inaccurate, you can exclude him for your Trust list already.
If DT1-members include the wrong users, they're supposed to be excluded. It's drama, but it comes with a decentralized system.

Letting jvanname drop distrusts for poor reasons or no reasons weakens the system.

He's not the only one, and I wouldn't say it weakens the system. It's just the result of not moderating feedback.

Ah we both agree it is not ideal, but  the question is how to correct it.

Obviously my distrust of someone would carry more weight than his as I simply do not do a distrust or a flag with out very careful consideration of why it should be done.

I also feel most DT1 people are the same way. ( note most not all )

If a systematic method of checking poor distrust use is developed I would certainly want to study it. Maybe try it for 1-2-3 months see if it works. tweak it etc.

[1miau]

Basing on experience with flags I can say that only minority of DT will vote for and against others' tags.

Yes and it's working pretty well for flags currently.  

Flags have rather strict rules and feedback doesn't. Otherwise there was no need in flags, it could be done on a base of quantity of tags. Some tags are left on the very boundaries of a consensus of a Trust system; flags in the same cases would have been opposed, because of strict rules.

I can't see how strict rules / not strict rules is making a difference. For example DT1 voting has no strict rules as well and bitching about DT1 inclusions / exclusions isn't a big problem at all.

If to make the same system as flags, then what will be the difference, why do we need two similar systems?

Because voting out certain trust feedbacks is a completely different thing compared to flags because it's serving a completely different purpose.  

Reviews becoming more and more flickering... it could destroy Trust system reputation.

Seriously, where should it "destroy" the trust system?
DT1 is currently determined that way, our own trust scores change by a small margin every week, depending on who's DT1 and DT2 and even DT2 is changing due to that already.
Nothing got destroyed from that.  

I said "destroy Trust system reputation", not the system itself. It doesn't matter if you are in DT1 or in DT2, your tags will stay visible by default the same way. If the number of votes for and against tags will change each DT1 reshuffle, there will be flickering in feedback: some tags will appear while some will disappear.

It's neither destroying the trust system itself and not even "destroying Trust system reputation". I don't see, where anything is "destroyed". Flags aren't destroying anything as well.

Does it look reasonable and reputable when the same tag of a same person is correct when there's one DT1 composition and becomes totally incorrect with another DT1 composition in a month and the same way each month?

Yes, it's completely reasonable because right now, we have also DT1 and DT2 changing frequently and therefore also trust feedbacks changing frequently. It's currently like that due to DT1 changes: some trust feedbacks are displayed this month and next month, they are not displayed, it's no issue at all.

No need for you to investigate such a case if you don't like to investigate. It'll be similar to flag support and it's working well currently.

Right, no need for me, no need for someone else. So if someone will gather with several like-minded they can change a system as they will like, because majority won't get much additional time to do much additional work to monitor the correctness of Trust system work.

Same issue goes for DT2 manipulation and there, it's a much bigger issue because a DT1 member can add a DT2 member very easily by just including the account. Manipulation, where abusers would need at least 5 DT1 accounts is much more difficult than to manipulate DT2 inclusion, where only 1 DT1 member is necessary.
You can be sure that if there's possible abuse, more DT1 members will look into it and overturn any abuser's votes.

As for flag system: it is different. Tags are much less strict. I voted for 205 flags because it is easy in many cases, but I left just a 100 tags and big number of them is based on one big farm investigation.

It doesn't make any difference as we know from DT1 voting, where it's also less strict and it's not an issue if it's not strict.
It basically boils down to what LoyceV wrote here:

We would then enter an endless loop of upvoting and downvoting. Basically, a popularity contest of who gets more DT1s to agree with them in a given cycle. The decisions made this month could simply be cancelled out next month or the month after that if the right or wrong people (depending on how you look at it) make it into DT1.

That sums up the entire DT1-system

[DVlog]

Same issue goes for DT2 manipulation and there, it's a much bigger issue because a DT1 member can add a DT2 member very easily by just including the account. Manipulation, where abusers would need at least 5 DT1 accounts is much more difficult than to manipulate DT2 inclusion, where only 1 DT1 member is necessary.
You can be sure that if there's possible abuse, more DT1 members will look into it and overturn any abuser's votes.

Why rely on votes from DT1 members only when there is a concern that all of them will not vote and many of them don't care? Can't we do a weight-based voting system? DT1 members will have 3 voting power and DT2 members will have 1. To pass a decision at least 50% of DT1 member's vote should be mandatory. This can reduce potential bias as well.

[decodx]

To pass a decision at least 50% of DT1 member's vote should be mandatory. This can reduce potential bias as well.

50% of all DT1 members? That will almost never happen.

[1miau]

Same issue goes for DT2 manipulation and there, it's a much bigger issue because a DT1 member can add a DT2 member very easily by just including the account. Manipulation, where abusers would need at least 5 DT1 accounts is much more difficult than to manipulate DT2 inclusion, where only 1 DT1 member is necessary.
You can be sure that if there's possible abuse, more DT1 members will look into it and overturn any abuser's votes.

Why rely on votes from DT1 members only when there is a concern that all of them will not vote and many of them don't care?

I don't see any issue if there are not enough DT members for some cases. The feedback will just stay as it is, so even for these cases, nothing is changing from what it is right now.
And for most cases, I'm sure a sufficient number of DT members will be attracted to look into it.

Can't we do a weight-based voting system? DT1 members will have 3 voting power and DT2 members will have 1. To pass a decision at least 50% of DT1 member's vote should be mandatory. This can reduce potential bias as well.

I agree, that a weight-based voting system is an interesting idea but from an implementation viewpoint, it might be more complicated. Probably, keep it stupid, simple is a better approach here, also to explain it to more DT members easily. Sometimes, I'm still of the impression that the whole DT system is very complicated, for quite a few long-time forum members.
Still, your idea of an weight-based voting system is definitely a suggestion to be considered. 

[Upgrade00]

Why rely on votes from DT1 members only when there is a concern that all of them will not vote and many of them don't care? Can't we do a weight-based voting system? DT1 members will have 3 voting power and DT2 members will have 1. To pass a decision at least 50% of DT1 member's vote should be mandatory. This can reduce potential bias as well.

That is pretty much still relying on votes from DT1 members and if you are positive to get 50% of them to chime into different discussions, how do you say that they do not care?

Your idea is one that can work and also help to reduce bias even though many believe that the default trust is already a broken system and encourages backscratching.

[Pmalek]

Trying to follow the alternating proposals here but failing to understand the logic of 2/3 or 3/4 of DT1 to vote. For starters, it's very unlikely to get 61 to 69 DT1 members to vote on anything imo.

I don't see the issue with having only 50%+ agreement as the threshold for having feedback removed...

It will undoubtedly be difficult, yes. I just feel like it shouldn't be easy to have feedback removed and put in the untrusted category. You are protecting the trust system that way. It already has its difficulties. If it was an easy thing to do, it would stop being a valuable tool and slowly lose its purpose. Maybe I am exaggerating.

look at my trust

https://bitcointalk.org/index.php?action=trust;u=64507

Hahaha. Sorry, had to laugh. If I had the power to make decisions here, such uses of the feedback system wouldn't be allowed. You are not the only that has that, and information that someone is stupid/crazy/gay/brain-damaged doesn't help anyone.

[dragonvslinux]

Trying to follow the alternating proposals here but failing to understand the logic of 2/3 or 3/4 of DT1 to vote. For starters, it's very unlikely to get 61 to 69 DT1 members to vote on anything imo.

I don't see the issue with having only 50%+ agreement as the threshold for having feedback removed...

It will undoubtedly be difficult, yes. I just feel like it shouldn't be easy to have feedback removed and put in the untrusted category. You are protecting the trust system that way. It already has its difficulties. If it was an easy thing to do, it would stop being a valuable tool and slowly lose its purpose. Maybe I am exaggerating.

I see your point that it wouldn't be ideal making it easy for feedback to be removed/un-trusted and you're probably right overall. For example a DT member could leave feedback that the majority of DT1 members agree with, but get's down-voted by a handful of DT1 and then removed. This is why I said having down-voting won't work without up-voting as well to counter it. The frustrating reality would be the possibility of having hundreds of pieces of feedback down-voted by certain members, that would require up-voting by others in order to remain, which sounds like an exhausting task for DT1 members.

Either way I don't think the feature should be rolled out without some sort of test-run first, for example a handful of feedback (proposed by users) that become available to vote on by DT1/DT members. Regardless of whether it requires 5 DT1 downvotes or otherwise a majority, I can see it going wrong either way, or being unpopular. Then ideally DT1/DT members would be able to vote on implementing the system or not.

edit: Am starting to think that this feature would work better by requiring a minimum number of users to initially propose voting on pieces of feedback, such as 3 DT members, before having the feature becoming available for the feedback in question - ie regulate the feature better. The idea of having every single piece of negative feedback available for up-voting/down-voting is starting to sound a bit insane in hindsight.