Thanks! Yes, I understand there to be a mix of people voting along with whoever is popular or just because they know more about them due to increased exposure - plus those who deliberately vote for others. We have enough votes that it makes sense for us to deliberately pick some 'big names' and also some who are less well known for our own reasons.
There are definitely some who will drop votes for the top 20 when they upset with Hard Forks' etc. too, yes.
It might be possible to psychologically profile the Pumpkin through statistical analysis of personality traits by comparison to the rest of the community, but I am not that obsessive :)
I have just spent a couple of days coding a page that demonstrates what would happen to the witness top 100 if we introduced witness vote decay (of 3 months currently in my tests) as I proposed months ago and which @ned appeared to like at that time.. From my initial data it really shakes things up dramatically as most of the top witnesses got their votes a long time ago, but we cannot accurately predict who would and would not reinstate their votes once they expired due to their time limit. If you are interested I can tag you when I make the finished page live.. I have some other ideas too which I will explain then which might make the situation more attractive to the average user and to investors in the blockchain too.
Vote decay has been suggested a number of times (at least in BitShares voting), but there are pros and cons.
The best argument for it I think is "dead votes" from accounts which no one controls any longer (e.g. keys were lost accidentally). In the long term, I think there will be a need for a solution like decay to weed out such dead votes that might be voting for no longer active (or even potentially rogue) witnesses, but there's a lot of different ways the decay could be implemented.
The biggest argument I can think against vote decay is that it could increase security risks associated with the same attack I mentioned before (someone hacks an exchange and gets a lot of "free" steem). Depending on the algorithm used for the decay, overall vote totals for witnesses could be lower, resulting in an easier attack based on large stake.
And, of course, depending on the actual decay algorithm chosen, there could be hassles for voters associated with re-voting.
Still, ultimately I think a voting decay algorithm of some sort is an eventual necessity. But I don't think it's a pressing concern yet.
Yes, the removal of dead votes is the main motive here, since the worst case scenario is that whales make big votes and then literally die and the votes stay forever, regardless of what the witness does later on.
I see what you are saying about the potential hacking of exchanges and stolen steem, but does that actually happen very often? It's not something I hear of much, but then I know that banks and perhaps also exchanges tend not to let anyone know when they are hacked to avoid harming their PR image. If large votes appear for witnesses from a hack, the results could still be a problem without vote decay, depending on the amount of steem stolen. Your point also raises an issue I hadn't thought of - what happens if a Steem account is hacked and cannot be retrieved for some reason? I guess there's no mechanism for blocking witness votes from that account.
I think if hackers can break into exchanges and steal steem, then they can steal any other currency too and then covert it into steem and no-one would really know publicly, so I'm not convinced that this risk is one that should stop the implementation of a well thought out vote decay feature.
I see this working well in conjunction with the idea of significantly lowering the number of votes that each account has. This would open the space up for more say from the wider community and also less risk of attack from stolen steem.
In any case, the page I have coded to display outcomes of vote decay rules is partially finished - I just need to do some bugfixing and test it some more. :)
It's certainly possible an exchange could get hacked and all it's steem stolen (unfortunately this kind of hack predictably seems to happen somewhere a few times a year, but no major amounts of Steem stolen to my knowledge).
Probably the best argument against such an attack is that most attackers would probably try to just sell the Steem instead of attacking the chain with it, as attacking the chain would likely just lower the value of the coin they sold. So it would probably take an attacker with something other than financial motivations.
I didn't mean to imply you shouldn't work on this issue, I'm glad to hear it. Mostly I was explaining why I doubt it was on Steemit's immediate roadmap.
As a side note, I just saw abit re-opened this issue on BitShares github repo and I heard recent discussion of it at the BitShares conference, so you might want to engage with some of the developers there as well.
I see, ok - I can appreciate why this would not be on Steemit inc.'s to-do list, in that they already have more fundamental issues to deal with - I agree. However, the recent complaints from many directions regarding the potentially limited scope of testing that witnesses engaged in prior to the last hard fork has nudged me to focus on this a while.. It is something that many have asked me to push forward anyway, but it's actually a low priority for me too most of the time too as I am pretty busy. I don't think many people really truly know exactly what was and wasn't tested - or what actually could have been tested, but that doesn't stop them complaining.
I think/feel that many people need to see some kind of steps being taken to possibly improve the things they complain about - whether or not they actually succeed is a secondary issue. The biggest challenge for Steem that I see in this regard is that among those who do deeper research, it is well known that generally speaking our mainstream 'democracies' really are nothing of the sort and are totally rigged/controlled. Since we see so many facets of government and wider society reflected in the structure of Steem and its users, the people who come here - attracted by the censorship resistant features and potential for causing disruptive evolution in society - tend to want to make sure that they are doing what they feel they need to do to produce a better outcome here than we see in the wider world in general. They feel powerless in mainstream politics and have given up there, so want to create something tangible here that is an improvement. Hopefully changes to the witness voting mechanism that are more competitive and enforce some degree of presence and activity on the blockchain in order to sustain a presence here, will go some way to increasing engagement and attentiveness all around.
In any case, I aim to open up conversation about how everyone might meet in the middle here in ways that benefit everyone.
Thanks for the tip about Bitshares, I'll see if I can find the issue you are referring to in github.
FYI, I just posted the initial data from my vote decay calculations (which doubles as a post showing the 'hottest' witnesses) - it's here if you are interested. :)