Bot Warz!!! A hybrid approach
All is not quite well, our problem looks a lot like this...
I'm a futurist and a transhumanist, I'm the last person in the world you will see actually complain about bots.
As a daytrader and programmer, I write them for myself thereby allowing me a hands free income. I also make a healthy living selling custom bots to other traders who just need an edge in the high stakes world of crypto trading.
Bots are nothing more than a force multiplier.
The truth is that there are tasks you can do around here that make money.
The more efficiently you can do those tasks the more money you will make. So you can't really fault the bots or their owners as long as they are still playing by the rules.
You can make money by posting content, upvoting the right posts (curation) and in general participating in the community.
As it turns out, posts featuring pretty girls, so called "cat fishing" and also posts welcoming new users tend to be the most lucrative, because our current demographics are what they are. Ergo, current generation bots are programmed to curate content targeting more of this. Unfortunately these activities raise the noise floor and make it difficult to find truly provocative content.
It's a negative cycle and I aim to do something about it.
This noise floor is the real problem here, not the bots.
If someone drops by and upvotes one of my posts I'm thrilled whether or not every vote I get is from a human. However there are bots which are randomly flagging content and downvoting perfectly good content, in addition to the upvoting bots that are not necessarily selecting for good content.
What we need here is to apply some survival of the fittest.
We want to starve the bots that are encouraging bad content and feed the bots that do a good job.
The noise floor issue is very similar to the problem of ads on websites.
The solution looks about the same as well.
Ads have pretty much been solved now via plugins such as adblock plus. We what need to build is a similar hybrid "bot/human" plugin that is capable of detecting, filtering and decentivizing bad content, while doing the opposite for quality content.
So here is the proposal.
Build a tool similar to adblock plus, but intended for steem content.
It will do the following items...
#1 It will watch your activity stream for you to flag or downvote content. When you flag or downvote content it will send an alert to other users of the tool indicating that you have done so. They will have the option of trusting your activity, thereby repeating it, or going to the content for manual review where they can flag or unflag and send the same message to their own followers.
#2 It will provide you with a personalized feed based on the activities of users who have upvoted content that you too have upvoted, which we take as a sign you are socially similar. There will be no reporting or data mining to do this, it will merely check the blockchain for posts you have upvoted and find the top 10 or so earners who are most similar to you socially by checking on content voted on in the past. All of this is blockchain based and it's all local to you.
#3 It will allow you to manually add & remove people on either list in case you disagree with your personal bot's findings or find others who interest you as well.
#4 It will provide an "autopilot" mode you can enable or disable at will. This way if you leave your computer for a few hours it can automate these activities so you can still contribute, even if by proxy.
I will begin work as soon as this posting hits $500 USD, and will post the phases publically for testing in subsequent posts about once a week.
If by some chance this posting breaks $10,000 USD I will also add a final feature that will allow you to "steem" as you surf. Highlighting and selecting content, putting some thoughts to it and sharing it on steem where the rest of us can enjoy it.
I believe that by leveraging the power of a human/hybrid approach we can have the best of all possible worlds. High quality content that targets our real interests; A healthy stable community and currency; The most intelligent bots.
What do you think? Would you like to see the power returned to the people before it's too late? Or is there a better approach to solving botspam than becoming a hybrid?
What you propose is similar in a way to a social trading platform mixed with a statistical matching / recommendation system. That's a really interesting idea which I think would really help improving quality and widening the focus of curating. I read that post too late to upvote it at the right time, and it seems that it hasn't gathered much interest. Don't let that deter your from developing your idea. Right now the voting system is completely broken, there is already way too much content for whales to look at, and the community has converged onto a bandwagon type of consensus around tabloid-like content as a local payout-over-timespent optimum. I believe (I hope) that we will manage to find the right balance after enough whales have down powered and sold some of their stake, and once we have changed the payout algorithm to grow more smoothly. But it could take a year before payouts play correctly their roles and posts like this one get the attention they deserve. Since you are a dev, I recommend you join the Steem slack where you will get a bit more attention. I'm interested to hear about the progress of this project: ping me on slack @talerecursion when you have some updates.
@pharesim @anyx are doing bots to help scale this. Personally I think all new users with no reputation should hit a quarantine phase. They get unquarantined after providing quality posts.
This may not be the best solution but until we have better evovled bots (for spam and plagiarism) it incentivizes users to provide insightful and original content.
While what they are doing is superficially similar, what is discussed here is a completely different approach.
You'll notice that both @pharesim & @anyx have proposals that rely on a central management scheme.
At some point there is an overseer who maintains a central list, and a single central "bot" that has a lot of power to up or downvote mostly based on how people up or downvote the bot.
Yeah that system makes sense but it will fail because at some point you have a single bot and intentions aside, that's giving censorship power to a central authority.
My proposal is the opposite of that. No bots doing anything. You flag a post and people who generally vote with you are notified about the action. They can choose to enable automation and just do as you do, or they can examine it. Eitherway it's just an autopilot that says "Hey @repholder I think that heroin junkie post at ... isn't legit. I've flagged it." or even "Hey @repholder , I just found a really neat post at ... I upvoted it, check it out and see if you can give it some love too."
There's no "list" per se. You see a bot, you click a button and bam all like minded individuals are notified of the opportunity to downvote or flag.
This gives us back our power collectively through the power of synchronized communication using nothing more than the blockchain itself. While negating to some degree the power and influence of the whales and the bots and even whale bots like @Wang (which has been caught upvoting for only a couple of people's posts regularly)
Does that make more sense?
I like the idea, it's something similar to what we are doing in steemitabuse. A bad actor posts something, someone notifies us of it and we go flag them. Sadly though these kind of systems tend not to scale as flagging causes vote power dilution (even for bots), people tend to work on what makes them rewards not something that would dilute future action.
Whales tend to feed the bots so no human intervention is needed, as they don't always have the time.
We've discussed this already somewhere. I think you should do it, but you should keep it closed-source to prevent it from being used by too many people. Here's why: suppose a bot is designed to learn what posts garner the biggest rewards. If one person runs this bot, they'll profit and promote good content, and that's a good thing. However, if everybody runs this bot, they'll profit, but they'll lose any likelihood of promoting good content. This is because the feedback loop will have closed - the bot won't be measuring what's good, it will be measuring what it thinks is good - and that could lead us off in a totally bizarre unexpected direction.
This is fantastic! But I totally disagree with you. First off, we all benefit from open source. Secondly I wouldn't run a closed source crypto anything. I trust myself to vet code, but I don't expect anyone else to trust me.
More importantly, we aren't talking about an opt out system here, it's opt in. You like a thing, it will find others who like similar things and add them to a list. You approve who is on your list though, so if you like the same stuff, you are notified when he/she likes something.
There is one problem with this approach that I see and it's the promotion of a monoculture.
If all you ever see are things you already like, if all you ever hear from are people you already agree with, then you become insular, isolated and tribal. As opposed to engaging in lively debate with mutual respect with people of dissenting opinions. Which for me is the fun of places like this.
I'm not sure how that could be solved to be honest with you.
Damn. Twice now I've tried typing a reply and it's vanished.
So carry on with your bot, I support it.
And if I haven't already pushed my game theory series on you, I will do so now.
Excellent articles I hadn't noticed them before now. But thank you, you are one of the most underrated creators on here.
the problem is, that even a bot could multiply his vote by voting at the right timing...
thanks william for voting my post
https://steemit.com/steemit/@knozaki2015/my-upvote-was-worth-5-06-usd-25300-or-how-to-earn-the-maximum-curation-reward
@knozaki2015 I'm not sure I follow you on that. Can you elaborate further please?
What a great idea! If you can actually program this, then you deserve minimum 500 bucks. I support you, mate. Weenis is getting on my nerves :)
Thanks! I'm working on the framework right now. There may be a bit of a curve getting it going. The best way for this to work is as a browser plugin. That would be fine except Chrome doesn't allow you to run plugins unless they come from the app store, even if you compile and sign them yourself.
Working on a solution to that right problem now.
is it going to be available for safari as well?
I hate bots in general but I love your idea!
Awesome! Glad you like this idea. I'll be making a new posting on the topic when I have the basic architecture nailed down. Hope you'll join us!
Checking my emails and posting a few thoughts here is about all I can do these days before I want to huck the laptop at the nearest hard spot. I use this internet thing for trying to find tribe and to post little ideas. I used to be big into computers but they just make my head hurt these days. I'll leave developing new stuff to people that are better at it and support those that are doing neat things.
This site is really heavyweight isn't it? Every time I load it I watch all 4 cores start to sweat.
Maybe the bot should just run standalone and snapshot and highlight relevant content you might be interested in, i.e. find tribe and post ideas.
I like the concept of a bot that can suggest and curate, but fear going too far towards clippyhell. OTOH any AI either needs to be rules based or super lightweight or it's going to melt your laptop. So maybe just a simple algorithm that can learn by adjusting it's weights based on activity.
That's all way over my head but this site sure does cook my computer and tear through bandwidth. Makes me wonder if something nefarious is going on. I hit inspect element on something yesterday and the computer crashed trying to load the code.
All I know is that seeing the term "AI" immediately gets my brain thinking about building a lipo based emp or herf gun. If good AI can kill the bad Ai, that may be our only hope though!