Holding back the spam tide by fighting automation with automation.
@timcliff posted earlier today, suggesting the possibility of a minimum payout threshold for comments as an anti-spam measure.
It wasn't wildly popular amongst those responding, but there's definitely a problem which needs to be addressed and I'm glad he's put it on the table.
In Tim's words;
One thing that I think increases the amount of spam is the fact that there is a small chance of earning rewards - even if the posts/comments suck. If a user posts 10,000 "nice post" comments in a month, and 10% of them earn a few pennies worth of rewards - then that is still a profitable business model.
Or as @tarazkp wrote in response;
This is the catch-22 problem isn't it? A cost can't be introduced because it will decrease interest levels in users but, without a cost, there is only upside to behaving badly. People want to have the best of both worlds but there is an incompatibility between the two. People want to protect the new accounts from costs so they can grow but in doing so means the system is getting swamped by those that only care for profit.
The first problem with many proposed solutions is that they need to be endorsed by the witnesses and go on to apply to the entire blockchain; the second is that they really struggle to disincentivise spammers without also punishing legitimate minnows.
What if blog pages had a new display telling us about their pending payouts and number of active posts and comments, with a "Downvote account" button?
You could catch me spamming, click my account name to go to my blog page, discern (according to your own subjective judgement) that you're looking at a spammer and choose to downvote the whole account, oldest active posts and comments first.
Your front end of choice, Steemit.com, busy.org etc, would calculate the total downvote strength required to drop each comment to zero, and downvote as many of them as your chosen voting strength would permit, starting with those closest to payout, working back to the present. Effectively voting many times with a single click.
You could then see the pending payouts and wait times change.
You could downvote the account just the once, or hit it over again until there's nothing left to land in their wallet.
You could dedicate one full strength upvote each day to the same account until they change their ways, knowing that they won't make a sneaky cent, while only having to check on the account every five or six days.
I recognise that this would make it slightly easier for vindictive flaggers to target legitimate accounts; however legitimate accounts can usually be flagged down to zero in a few minutes anyway since they tend to have fewer, larger paying posts.
Unless I'm mistaken, this could be handled by any app in possession of one's posting key, and would give the anti-spammers a similar level of automation as that enjoyed by their quarry.
The real problem we have is that those without money but with lots of time, are able to use simple bots to spam comments; while those with the voting power to tackle them and with little time to do so, are trying to downvote these comments manually.
Our issue isn't with a particular piece of content, but with the activities of the account overall; so why require that those looking to fight this scourge, scroll through thousands of spammy comments hoping to find some which haven't yet paid out?
Pitfalls, possibilities? Is this not technically feasible? Let me know in the comments.
As always, have a fantastic day.
NB: I recently posted a Dlive clip discussing quadratic vs linear rewards; but a lot of people advised of excessive buffering so I've reposted it to youtube - here
I've never had trouble finding and/or identifying spammers. The big rings of them have thousands of accounts. Automating it at an account level wouldn't be enough - I'd need something I could put a long list into. (Of course I'd also need the SP to make my vote able to nuke 7000 spam accounts, but you get the idea).
I'm also with other commenters such as @tarazkp in voicing a note of caution on the potential for this to be abused. Having taken a few flags this week over ideological disagreements about bidbots and what constitutes spam etc., I'd be cautious about making it too easy to zero an account. Maybe run this idea past @freezepeach et al for their perspective.
Wololo!
(RandoHealer has healed this post because your blog was targeted for healing due to malicious downvotes!)
So it's simultaneously too powerful, and not powerful enough
Sounds like a happy compromise :)
Any feedback from anyone welcome.
@arcange, @steemcleaners; would this make it easier for you, or too easy for malicious flaggers?
People already have the ability to find spammers by analyzing the data. Such Spam Comments like that post a couple of months back. Not to mention there accounts out there that do go after largescale spam that are using upvote circles to profit. Spaminator is one account that comes to mind.
If just downvoting the spammers were good enough they be taken care of already. People in the past have made daily hit lists to go after people spamming. They can just make hundreds of new accounts at no cost to them and that is the issue. It’s a flood gate and while it remains open its just an annoying mess to deal with any.
Last week I had 132 comments with 63,766 characters total. Some users can do 300 to 600 comments in a week just enjoying steemit without running a spam bot. Someone could easily go oh 600 open comments must be a bot spamming and nuke them.
Now can I 600 comments in a many in a week? Heck no but I also take much longer to write comments and I write blogs. There are some legit users who only leave comments and never blog.
That's why a completely automated solution is going to pose problems.
We need people to manually decide that an account is spamming; this just speeds up the downvoting.
Did I just find a Facebook full of people with interesting things to say?
I am going to have to poke around on github. Y'all have got me quite curious how this reward system works exactly. I get that there is steem and staked steem that becomes steem power giving you more clout (plus the POS) but I did not realize or think about the effects of factors such as frequency, history and observed future; how that can effect ever up or down vote ahead. I am just going to have to dig as I have too many question to not be a pest. Really I have not even gave the FAQ a good look yet. Thanks for sparking my curiosity guys.
We're still in beta, so clearly still ironing out the kinks :)
Sing out if you're struggling.
So much brainstorming. I hope a workable solution comes up. The main problem I see with solutions that make it easier to downvote spam is that it also makes it easier for malicious downvoting, which seems to be getting more and more common. Complain about my huge money making bidbot? Goodbye post... Point out my hypocrisy? Goodbye post... Grumpy animals admitting to censorship... You get the picture, I'm sure. The ones with the most money = power in the governance.
Can anarchy ever exist with the drive people have for power? If you remove the ability for any one man to have so much power that they have control over others, then is it still within the realms of anarchy? Now I'm just digressing though.
Anarchy just means that the property owner sets the rules.
In that case, we're all guests at Ned's house.
That could open a can of worms! As far as they're concerned, government (or Crown Corporation) owns all the land, so are we guests at their house?
No. They didn't mix their labour with nature in order to homestead the land.
They waved their hand over an entire continent and said, "mine".
That's not how ownership works.
Good answer!
This I think would save hours in @berniesanders' day :)
I think that this would make it easy when nuking a spammer but, also easy for people to abuse in regards to complaints about rewards. Will they read the content or even scroll through the account to see what is really there or, just nuke? If you or I are going to nuke, we are likely to d a little due diligence first but many may not, and they will use it for their vendettas.
I enjoyed your vid on the voting types. BTW it loaded and played fine on DLive. Well. done.
Thanks mate. A lot of new faces hearing phrases they don't recognise, helps keep everyone on the same page.
An interesting idea with the downvote button I do not know. It would have to be tried. I tend to ignore all the spam. Thanks for thinking outside the box and you continued efforts to fight this spam @mattclarke
Definitely sounds feasible to me if there is demand to use it and develop it. I wouldn't be looking to Steemit for that though. Maybe more like a SteemVoter type of service but for auto-flagging?
I've run into a few situations where my software would make too many calls to update (such as the invalidate procedure in the windows api which would cause flickering on the display if the drawing was shown in real time). This spamming of the OnUpdate method pointer I used to handle by drawing on a hidden TImage object.
The takeaway from this is that if a routine starts to become a problem, if you want to discourage the behavior, you should minimize thread use. A damping algorithm can be used for successive iterations that reduces the profitability beyond a certain number of posts in a given period. Short spam posts are easy to spot and long spam posts can be handled similar to the way @cheetah works.
I'm not sure how many posts per day a typical legitimate user does, but I've been on here since late last summer and still haven't cleared the 1200 post mark. At 1170 posts in 218 days, that's about 5.37 posts per day. I'm probably on the low end of posting usage on Steemit, but I imagine that a damping algorithm can start coming into play at about 50 posts or more in a given 24 hour period that gets progressively onerous as you exceed that number of post effectively eliminating payouts at thresholds that spam bots use.
Some bots are worthwhile though.
Some warn people if a previous comment was a phishing link for example.
We need human eyes to sort the spam from the steak, but once we've identified it, we should have the tools to nuke it, ready to go.
Chips and tomato sauce...So good.
😂