BitShares Forum

Main => General Discussion => Topic started by: Agent86 on November 12, 2014, 04:05:24 pm

Title: Throttle delegate approval?
Post by: Agent86 on November 12, 2014, 04:05:24 pm
I want to see what people think of an idea to throttle approval of delegates.  The point is to allay fears that some entity accumulates a large secret stake and then quickly votes in otherwise unknown delegates before anyone has time to react.   Throttling could allow time to publicly demand an explanation if unknown delegates start getting huge support.  It would give time to "get out the vote" to coalesce around more well known delegates so the offending delegates don't get in or even prepare to consider the extreme measure of hard forking out stake if no one knows why these delegates have any support.  Basically there is no restriction on a delegate losing approval but a delegate can't gain more than X% approval in a 24 hr period or similar rule.  Kind of like the idea that trust takes time to gain but is quick to lose.
Title: Re: Throttle delegate approval?
Post by: matt608 on November 12, 2014, 04:24:20 pm
My thoughts on this:
https://bitsharestalk.org/index.php?topic=11241.msg148466#msg148466

TLDR -  Delegates become trusted with dilution ability (over their basic 3%) after a certain amount of time in the top 101.
Title: Re: Throttle delegate approval?
Post by: Rune on November 12, 2014, 04:28:34 pm
I think the 2 weeks registration fee solves this. The only risk of "instant delegates" would be if someone where to amass enough to perform a 51 delegate attack. I don't think that is a real risk, or if it happens we will have much recourse anyway. It would only happen due to some really skewed incentives, though, because someone with that amount of stake would take massive losses from attacking their own investment.
Title: Re: Throttle delegate approval?
Post by: Fox on November 12, 2014, 04:31:16 pm
The primary role of the delegate is to produce blocks containing valid transactions.  Any delegate failing to do so will be voted out.  I see little harm in allowing the network to operate as designed and continue to enable the shareholders to vote for delegates as rapidly as they see fit. 
Title: Re: Throttle delegate approval?
Post by: Agent86 on November 12, 2014, 05:00:02 pm
I think the 2 weeks registration fee solves this. The only risk of "instant delegates" would be if someone where to amass enough to perform a 51 delegate attack. I don't think that is a real risk, or if it happens we will have much recourse anyway. It would only happen due to some really skewed incentives, though, because someone with that amount of stake would take massive losses from attacking their own investment.

Yes, this proposal was to address the idea of a 51 delegate attack more than anything else so people would "see it coming" so to speak.  I think seeing it coming, and knowing that you would see it coming could have some benefit.  I don't think the 2 week registration fee does anything to help but I'm definitely not convinced this is a good idea or worth doing.  Just wanted to see people's thoughts.

My thoughts on this:
https://bitsharestalk.org/index.php?topic=11241.msg148466#msg148466

TLDR -  Delegates become trusted with dilution ability (over their basic 3%) after a certain amount of time in the top 101.
Dilution happens over time anyway so I wouldn't be in favor of a time limit specifically before dilution happens.
Title: Re: Throttle delegate approval?
Post by: fluxer555 on November 12, 2014, 05:10:54 pm
I think the 2 weeks registration fee solves this. The only risk of "instant delegates" would be if someone where to amass enough to perform a 51 delegate attack. I don't think that is a real risk, or if it happens we will have much recourse anyway. It would only happen due to some really skewed incentives, though, because someone with that amount of stake would take massive losses from attacking their own investment.

The registration fee is a very sloppy way of handling this. It may solve this issue, but the fee has other problems, which complicate the incentive structures for creating new (valid) delegates, and which unnecessarily makes running for office a very risky endeavor. I think we can do better than that.
Title: Re: Throttle delegate approval?
Post by: gamey on November 12, 2014, 05:19:28 pm
I think the 2 weeks registration fee solves this. The only risk of "instant delegates" would be if someone where to amass enough to perform a 51 delegate attack. I don't think that is a real risk, or if it happens we will have much recourse anyway. It would only happen due to some really skewed incentives, though, because someone with that amount of stake would take massive losses from attacking their own investment.

Yes, this proposal was to address the idea of a 51 delegate attack more than anything else so people would "see it coming" so to speak.  I think seeing it coming, and knowing that you would see it coming could have some benefit.  I don't think the 2 week registration fee does anything to help but I'm definitely not convinced this is a good idea or worth doing.  Just wanted to see people's thoughts.

My thoughts on this:
https://bitsharestalk.org/index.php?topic=11241.msg148466#msg148466

TLDR -  Delegates become trusted with dilution ability (over their basic 3%) after a certain amount of time in the top 101.
Dilution happens over time anyway so I wouldn't be in favor of a time limit specifically before dilution happens.

I think these are all steps in a positive direction.  The 2 week has some value if you don't consider the downside.

The value of 2 week rule will have more value when the market cap grows and full dilution is a rather large amount. Then we are going to have huge incentives for people to carry on scamming campaigns without the 2 week fee.  This has benefit of not just the chain losing money, but it prevents a certain element from coming around and making us look bad.

However -
My belief is if we end up losing diluted funds to bad actors then it will be a small price to pay for the increased quality.  We need to tie the burning action into the delegate voting screen, so that we have an average accepted income.  By and large people will be producing far more value than they would in a traditional corporation.  This excess value will pay for the times the chain gets scammed.

The value proposition for being a paid employee is really hurting right now.  In reality we may know how it will work and I realize I am being a bit dramatic in my concerns.  Yet I still feel they are valid.

I do think that this creeping delegate approval instead of 2 weeks is an exceptional idea.  The marketing from this as backup system would be great (to crypto-currency geeks).  If we ever have to rewind transactions, it will mean some people get screwed - this idea could really go a ways to lower the chance of this happening. You are taking a questionable idea and swapping it out with what seems like a great idea.  Thats not something we can do often.
Title: Re: Throttle delegate approval?
Post by: gamey on November 12, 2014, 06:49:44 pm

+5% Agent86.

Why does no one discuss this proposal and instead gets their panties tangled over Rune's idea?

Is it because the idea of taking over Bitcoin is so exciting?  Actual doable incremental improvement for BTS is not so interesting?
Title: Re: Throttle delegate approval?
Post by: hadrian on November 12, 2014, 09:09:22 pm
Sounds good to me.

I'm wondering if there any circumstances under which this proposal would be a hindrance.
Would it cause extra 'load' on the system programmatically?

I can envisage scenarios whereby it could be useful to have the ability to vote delegates in 'from nowhere' immediately. These scenarios, however, would only exist after BitShares had already morphed into a very different beast, meaning it's currently irrelevant.
Title: Re: Throttle delegate approval?
Post by: gamey on November 13, 2014, 03:10:04 am
Sounds good to me.

I'm wondering if there any circumstances under which this proposal would be a hindrance.
Would it cause extra 'load' on the system programmatically?

I can envisage scenarios whereby it could be useful to have the ability to vote delegates in 'from nowhere' immediately. These scenarios, however, would only exist after BitShares had already morphed into a very different beast, meaning it's currently irrelevant.

I couldn't think of a way this hurts BTS.  It also doesn't seem like it would be that hard to implement in code but that would take a developer familiar with how it is currently done.
Title: Re: Throttle delegate approval?
Post by: zerosum on November 13, 2014, 06:06:57 am
I like the idea with one condition:

we should have enough Throttle-out/vested delegates in positions 102-151....
So if we need to replace a bad actor(s) we do not have to wait 2 weeks for some new delegate to become Vested/Throttled.

[This might be addressed/obvious, excuse my post if this is the case] 
Title: Re: Throttle delegate approval?
Post by: gamey on November 13, 2014, 06:34:04 am
I like the idea with one condition:

we should have enough Throttle-out/vested delegates in positions 102-151....
So if we need to replace a bad actor(s) we do not have to wait 2 weeks for some new delegate to become Vested/Throttled.

[This might be addressed/obvious, excuse my post if this is the case]

I hadn't thought of this.  I kept thinking of delegates in terms of paying a salary.  This goes back to the problem with having employees and delegates being the same.

I assume that the uhh approval creep would work on those under 101 also, so that they would be quite close to being moved in when called upon.
Title: Re: Throttle delegate approval?
Post by: hadrian on November 13, 2014, 05:56:21 pm
I like the idea with one condition:

we should have enough Throttle-out/vested delegates in positions 102-151....
So if we need to replace a bad actor(s) we do not have to wait 2 weeks for some new delegate to become Vested/Throttled.

[This might be addressed/obvious, excuse my post if this is the case]

I was thinking about this also.
This part of the idea is relevant:
...Basically there is no restriction on a delegate losing approval but a delegate can't gain more than X% approval in a 24 hr period or similar rule.  Kind of like the idea that trust takes time to gain but is quick to lose.

So if a bad actor needed to be replaced, hopefully many of their votes would be removed by stakeholders, then those delegates next in line would automatically rise without any need for their % approval to rise significantly.

If i'm right  :P
Title: Re: Throttle delegate approval?
Post by: Method-X on November 13, 2014, 06:15:05 pm
Here's a concern that's been on my mind lately: if a shareholder with a large stake (let's say 7%) were to either die unexpectedly or lose access to his private keys, would the delegates he voted for have 7% approval forever? If true, I propose votes expire after a year.
Title: Re: Throttle delegate approval?
Post by: matt608 on November 13, 2014, 06:50:38 pm
Here's a concern that's been on my mind lately: if a shareholder with a large stake (let's say 7%) were to either die unexpectedly or lose access to his private keys, would the delegates he voted for have 7% approval forever? If true, I propose votes expire after a year.

Vote decay has been suggested here and it looks like it was declined by BM on the next page of the thread:
https://bitsharestalk.org/index.php?topic=10937.msg144860#msg144860
Title: Re: Throttle delegate approval?
Post by: Agent86 on November 13, 2014, 07:31:43 pm
Here's a concern that's been on my mind lately: if a shareholder with a large stake (let's say 7%) were to either die unexpectedly or lose access to his private keys, would the delegates he voted for have 7% approval forever? If true, I propose votes expire after a year.

Vote decay has been suggested here and it looks like it was declined by BM on the next page of the thread:
https://bitsharestalk.org/index.php?topic=10937.msg144860#msg144860

I think the general consensus has been that we assume people have lost their keys if they don't make a transaction at least once per year...  This was part of the idea behind the inactivity fee and I think it holds true for votes also.  I think BM opposed the "vote decay" mostly because of calculation expense but I don't think that means he opposes votes expire after one year.  I support votes expire after one year and stake is considered "inactive" if it hasn't moved in one year.