Steem Bounty System - Milestone #1 (Proposal)
I have been thinking a lot about bounties over the past few weeks, how to systemize it, and how to imbue the spirit of steem directly into it's core. This document focuses on technical requirements, systems and the user experience. I am not the best designer in the world, so I didn't take a stab at the ui/design aspects of the project. Those are secondary to the actual systems behind it anyways.
This is what I came up with. It's huge, so I broke it into 3 major parts:
- A technical/ux concept for an MVP bounty system
- Further analysis of specific concepts
- My personal thoughts on this system
If you're planning on diving into this entire post, I'd recommend setting aside an hour or two to fully comprehend it's design.
This post is in direct respond to Milestone #1 from @ned's post, It's time for a steem bounty system.
Part 1: The Bounty Website - MVP Edition
Terminology used in this document
Briefly, here's some of the terms I will be using throughout the document.
- Bounty Creator: The originating user who submits and funds the bounty
- Bounty Hunter: A user who is submitting work to claim the bounty
- Bounty Backers: A user who votes on the bounty announcement post
- Bounty Contributors: A user who contributes additional funds to the bounty
- Bounty Hunter Claim: Submitted work by a Bounty Hunter to claim a bounty.
- Bounty Reward: The total reward pool built around a bounty (funding, post rewards, etc).
- Bounty Website: The specific website that powers this system. I don't have a name for it, so I'll refer to it as such.
Theoretical requirements for submitting a Bounty in the MVP system
- A task that consists of a single phase & reward.
- A task that in which a reward will be given to a single Bounty Hunter.
- An initial funding of at least 10.000 SBD (arbitrary minimum amount).
The internal life cycle of a bounty
A bounty, as an entity, will have the following lifecycle to flow through. This concept should also be further elaborated into a flow chart with all labeled decisions.
- Bounty Submission
- Bounty Creation
- Bounty Hunter Claims
- Bounty Conclusion
a. Completion
b. Cancellation
c. Abandonment
1. Bounty Submission
The Bounty Submission phase is a process performed by someone who has initial funding and wishes to pay someone to complete a task. This person will be the Bounty Creator and associated as the primary point of contact behind this task.
- Bounty Creator visits the Bounty Website
- Clicks "Create New Bounty" (no registration necessary)
- fills out task title
- fills out an introduction/background for the task
- fills out the tasks that need to be completed
- fills out any additional requirements
- fills out contact information
- submits form
- Submission creates a database entry and a unique identifier for the bounty
- Bounty is not yet active
- Bounty now has a lifespan of 24 hours, in which unfunded bounties will be deleted
- User is redirected to a confirmation page
- User is given the opportunity to review/change the bounty before it's finalized
- User is given instructions on how to fund the bounty using SBD/STEEM
- Multiple options can exist here, steempay.io/shapeshift/cli_wallet commands
- Bounty Creator transfers initial funds to @bounty with the identifier provided in the memo field of the transaction.
- Upon receiving of initial bounty deposit (meeting the minimum bounty requirements):
- The first account to deposit is now associated to the bounty as the Bounty Creator
- Future versions of the Bounty Website could include controls for the Bounty Creator to manage the bounty into the future.
- The confirmation page dynamically updates and shows completed, with a link to the appropriate pages.
- Link to view bounty on the Bounty Website.
- Link to view the announcement post on steemit.com.
- Link to documentation on how you should promote your bounty.
- The bounty information can no longer be edited (also due to the 24h lock on posts).
- A future version of the Bounty Website could allow the Bounty Owner to "amend" the bounty, showing a full changelog of how the bounty has evolved.
- The first account to deposit is now associated to the bounty as the Bounty Creator
- The system moves onto the next phase, Bounty Creation.
2. Bounty Creation
Upon a successful funding and creation of a bounty, the @bounty bot will create a post on the blockchain with an announcement of the bounty.
- The @bounty bot will create a post:
a. Using the #bounties tag (or whatever tags deemed by the community)
b. The post will contain:- A summary of the bounty, as filled out in Phase #1
- A bit of information about the Bounty Creator
- A large prominent link to visit the Bounty Website to view the live status
- This will be important because after 24 hours, the information in the post will be locked
- Instructions for Bounty Hunters on how to submit a claim to this bounty (details below)
- Instructions for Bounty Contributors on how to add additional SBD to the reward pool (details below)
c. If the bounty has been funded with 100 SBD or more (arbitrary amount), the @bounty bot will also vote on the bounty post, contributing to the reward pool and visibility.
d. All SBD rewards from the post (24h + 30d) are automatically allocated to the reward pool for the bounty. - This allows a form of non-monetary contributions from the community to support
- Any post rewards will be distributed once they're available and the bounty is complete.
- The Bounty Website will begin displaying the bounty
a. All of the information submitted by the Bounty Creator will be displayed on the page of a specific bounty. More details about this page below.
b. The bounty will show up on the homepage and the pages which allow you to browse bounties.
3. Bounty Hunter Claims
For the MVP of this product, we would create Bounty Hunter Claims using the steemit.com posting interface. As the Bounty Website continues to evolve, this process could be internalized within the Bounty Website for ease of submission.
- Each bounty has a unique identifier associated to it, displayed prominently on the Bounty Website and on the steemit.com post. To submit a claim for the bounty, the user must:
- Create a new post using the
bounty-submission
tag as well as a tag with the unique ID of the project, e.g.bounty-57ccca17c0c84a002b704b71
. - This post will contain the required information to complete the bounty:
- Images if it's an image task
- Content if it's a content task
- Links to external files if it's a development task
- The Bounty Hunter could also earn a reward for writing this post and potential tips for the effort.
- Create a new post using the
- Each Bounty Hunter Claim submitted will be reviewed by the Bounty Creator.
- The Bounty Creator could be notified via email/slack/etc of new submissions
- The Bounty Creator will provide necessary feedback on entries as needed.
- All of the Bounty Hunter Claims that are tagged with the
bounty-{id}
tag will be aggregated and displayed on the Bounty Website as potential claims to the bounty.
4.a Bounty Completion
Once a suitable Bounty Hunter Claim has been submitted, and the Bounty Creator is satisfied. The Bounty Creator can signal the end/award of a bounty. For the MVP of this product, it will be done through the issuing of a specific command from the originating account.
In essence, we will have a single command that needs to be issued:
bounty-{id} award account_name
This command can be issued in a number of ways:
- Sending a transfer of 0.001 SBD to @bounty with the command as the memo.
- Replying to the bounty announcement post with the command as the body.
- A future private messaging system with the command.
- Creating a new post with the command as it's body and a specific tag.
Currently this system is designed in that the Bounty Creator picks an entry to award the bounty. Future "types" of bounties could be created that also allow for voting of the winner or multiple users deciding.
4.b Bounty Cancellation
If for any reason the bounty owner wants to cancel a bounty, the bounty owner can issue a special command (using the methods above) to halt a bounty. Bounties may need to be canceled due to lack of participation, or failing to reach a specified amount.
bounty-{id} cancel
Once a cancellation is started, it will enter a 3 day period in which funds are released from the time locked savings. During this time, a cancellation can be challenged by anyone to halt the withdraw and dispute it. A user may be in the middle of working on something and wants an opportunity to display their work. Human intervention and moderation will be required at this point for the MVP to ensure no abuse of the system is occurring.
After the 3 day cancellation process, all funding to this bounty will be distributed back to the Bounty Contributors who contributed at the exact amount they contributed. Any remaining SBD (from post earnings) will then be powered up by the @bounty bot.
4.c Bounty Abandonment
If a bounty is abandoned, currently the team behind the Bounty Website would have to reconcile things manually. Questions remain about what constitutes an "abandoned" bounty and efforts should be made to contact the Bounty Creator should this situation arise. Developing an automatic system for this aspect should wait until further evaluation is done.
Part 2: Further analysis and elaboration of specific concepts
Bounty Contributions
There are two ways anyone can contribute to a bounty and it's reward:
- Large impact: by contributing liquid SBD/STEEM into the bounty pool
- Small impact: upvoting the announcement post of the bounty
Both of these classes of users will be highlighted on the Bounty Website as users who are helping contribute to the success of the platform. This idea could then morph into leaderboards and profile pages, giving people a place to proudly show their support. Most of this concept should remain outside of the MVP of the product, but can easily be grown into.
Disputes
All disputes will be handled by humans in the beginning. Any part of transferring funds or awarding Bounty Hunters can be disputed and handled by hand.
Arbitration Council (potentially) - A team of volunteers could be assembled/voted on to handle all disputes that require human intervention.
Disputes and the 3 day saving account withdraw
The bot during it's distribution of rewards has a built-in 3 day dispute period, thanks to the time-locked savings accounts. During that 3-day window, any disputes can be surfaced and the transaction can be halted. This should provide a sufficient oversight window.
Multiphase bounties (aka milestones)
Multiphase bounties should be broken up into multiple bounties for the MVP of this product. The additional complexity it causes should be moved to a later date to avoid release of the initial product.
Promoted Posts
The promoted post system could be used to promote bounties to larger audiences. Anyone using the current system could promote a bounty if they choose to do so.
Multi-sig Bounty Creator
Bounty owners could technically be groups of users who are in control of a multi-sig account. A Bounty Creator could set their posting key to multi-sig to require the modification of a bounty.
Community voting for bounty awarding
This currently can't work, there's a number of reasons as to why we shouldn't use voting on posts/comments as a mechanism to choose who to reward. This could be an entire post about the mechanics of voting and why it wouldn't be the best fit.
Multiple Bounty Hunter Rewards - reward splitting
In an initial capacity, this should be considered a "nice to have". It creates additional layers of complexity in the automation of rewards distribution, and because of this, shouldn't be considered for the MVP.
In the future, being able to state "3 rewards" during the creation of our bounty would be ideal. This would then cause the system to split the rewards 3 ways based on the 3 accounts the Bounty Creator has chosen to award.
Bounty Time Limits
The concept of time limits on bounties should be explored outside of the MVP. Determining appropriate lengths and the parameters for abandoned projects will be learned over time.
Transfers and Testing
To start out with, I'd recommend that all transfers actually use multi-sig and someone overseeing the program day to day would ensure the bot is distributing funds correctly. I doubt the volume would be high enough to make this an incredibly burden, but people would need to know that there's a human blocking some transactions so it may take days for approval.
Weekly Bounty Recap
Each week, the @bounty bot could create a recap post of all active and recently completed bounties. This would serve as a health-check to the community and as additional visibility about active bounties to the community.
It's another method for the @bounty bot to grow in power and use it's influence to promote bounties.
- All active bounties will be displayed in the post with their current information
- All bounties completed in the last week will then be listed with results
- The SBD earnings of this post could be redistributed into all active bounties or used to fund development of the service
- Encourages new users to submit bounties + votes on active bounties
@bounty bot concept
- Acting as a "whale"
- The bot could act as a sort of whale account to help bounties with visibility and rewards. Specific criteria would likely have to be enforced to disallow gaming of it's power.
- The bot itself grows in power based on SP rewards by it's posting and curating activity.
- It could be used to "tip" people who submit meaningful, but not finalized work to a bounty.
- Helping self-fund it's own development - paying for infrastructure and development costs.
- The bot could author non-bounty posts (maybe patch notes for the program?) to earn SBD + STEEM, then used to pay for services.
- The bot could be powered down for a weekly contribution to costs.
@bounty bot behavior
- Incoming Transfer Operations
- Any transfer sent with a matching bounty ID will be automatically pooled with other matching IDs
- Any transfer sent that has an invalid memo will automatically be send back to it's source.
- Outgoing Transfers
- Once the specific conditions (smartcontract someday?) are met, the bot distributes the awards accordingly.
- Time Locked Savings
- Potential Escrow Usage
- It's possible the bot could use the new escrow feature in some way, but it yet unknown. Currently with the crowdfunding nature of this proposal, escrow may or may not make sense.
Cost of maintaining the service
- The initial development of this project isn't a small feat, especially when you're talking about holding and escrowing funds for projects. Security and monitoring should be considered in every aspect of the development of this system's automation.
- I would put roughly 2-3 months of development time on the MVP of this project to build the foundation right. It may go a lot faster than that, but it's a comfortable timeframe to not feel rushed and cut corners.
- High level components:
- The frontend Bounty Website
- Submission of Bounty
- Viewing of a Bounty
- Browsing, searching, filtering of Bounties
- The @bounty bot(s)
- A process to monitor and handle automated transactions
- A process to create and monitor posts
- Support System setup and considerations
- Infrastructure Setup
- The frontend Bounty Website
- Ongoing/future development
- Maintaining APIs with the changing steem blockchain
- Implementing native solutions and creating a single site experience
- Integrating new payment methods for funding
- Adding user/management interfaces
- Mobile implementation
- Monitoring
- Post monitoring for spam/abuse
- Daily Reports to keep tabs on if any attempts to game the system occur
- Vigilant monitoring of the bots and wallets
- Infrastructure
- This setup requires a multi-server infrastructure. Estimating ~100 USD a month.
Part 3: My personal thoughts on the bounty system and going forward
TLDR: This project, done right, really isn't as simple as it sounds.
As you can see from the above outlines and thoughts, building a system that embraces the reward structure and ideology of steem isn't as simple as customizing a CMS. I've spent a non-insignificant amount of time already exploring the concepts behind it, and I didn't even scratch the surface of an effective user interface/design.
I am incredibly happy to share this thought experiment with all of you and hopefully some of the concepts and ideas will help spark new thoughts. There are a number of concepts here that could be expanded upon for the bounty system or potentially for other applications.
To close this out - I am uncertain currently of how much further I will be pushing this project in a development capacity. The bounty system is a large undertaking that I could see myself spending the next few months absolutely focused on. It's also a huge responsibility to take on, especially regarding the security and integrity surrounding the system. I'll be exploring this idea further over the coming week or so to really decide how to proceed.
did you see this?
http://bounty.steemr.io
Yep, I did. I actually left feedback for @steve-walschot in his post about it as well.
Great job @jesta! I really like the idea of running it through the @bounty account to handle the posting of the bounties. Its well thought out :)
I feel it gives a layer of trust that whoever's posting the bounty isn't just trying to scam rewards.
I also got lucky enough to snag the @bounty account, which I'm more than happy to surrender if there's traction by someone else on the project.
Awesome! I was wondering about that :)
I see a way of easy misuse.
This system, without 3rd party oversight could be abused. It can very easily be setup to conduct fraud, theft, Ponzi schemes, etc.
In general I like the concept of a bounty system. I think there needs to be more controls designed with malicious parties in mind.
Ah hah ha hah, So you actually saw my Save The Whales article. I assure you, I had no malicious intent.
Overall I see the things you speak of all being things not related to an actual "bounty" program, but more of a crowdfunding platform. Sure some of it may apply, but in these examples, I think the existing safeguards would work just fine.
These are also things you could do right now simply by writing posts. The system proposed above doesn't enable any of this behavior specifically, though it would present it differently.
With all due respect, don't underestimate the minds or influence of professional fraudsters. Just because you don't think they can do it, does not mean they can't find a way. Wetware (people) are much easier to hack than technology.
I agree this is tied to the weaknesses of crowdfunding, which in this case is very similar to the bounty proposal. But there are other aspects, social aspects, which could play into this. In fact, I blogged about some of those earlier today https://steemit.com/security/@mrosenquist/cryptocurrencies-are-a-target-for-cybercriminals-social-platforms-are-next-part-2
Don't get me wrong, I think a bounty program is a really good idea. I just want to make sure we have the appropriate controls in place to reduce the likelihood of abuse AND limit the impact when it does occur. I like an independent 3rd party who make decisions on allocations. The creator, is just that, only creates the bounty. They do not actually allocate the funds.
Taking a page from Steem, why not have a council of witnesses, including a number who are random, that must vote to disposition funds. This might be applicable for any bounties reaching $500 or more. They alone determine if the bounty has been fulfilled and allocate funds. The volunteers also get paid a little for the service. Such a system would be much more difficult for an attacker to setup as they don't control who the witnesses are and therefore can't control the destination of the money. Just an idea.
Apologies, I didn't mean to come off as dismissive at all. I actually wrote that comment in somewhat of a hurry (was leaving the office after a long day).
I agree that it shouldn't be underestimated, and I agree that a group should likely be performed to provide oversight (which is why I mentioned the council idea).
Ideally you're right, a 3rd party who makes decisions would be nice, but the technical complexities of designing the system or the manpower required to operate it are now 10 fold. It's not really a viable option if I were to build the system.
To counter this idea - I plan on using the bounty system create bounties for my projects. I don't want a council deciding anything, I want to be the one deciding. No offense, but you don't know what's best for my projects. If people don't trust me, then they shouldn't apply for my bounties.
I didn't dive into the community driven awarding aspect for these reasons - my feelings towards it are mixed and complicated. Specifically when I tried to break down to create engineering requirements, I couldn't find a system that wasn't a full time job for many people, that scaled, and/or wasn't abusable.
With the way that the simple dispute system written above works - anyone involved in the bounty can stop the automated process requiring manual intervention. This was the cleanest way I could systemize that actually worked.
I would love to see an idea otherwise though that didn't require a ton of effort on either development or the council :)
No apologies needed!!! This is a good discussion.
You bring up an interesting point. I doubt anyone, myself included, would question your bounties. You have contributed so much to Steemit! Your reputation precedes you. Perhaps that is another angle to explore, the reputation of authors. That could be a mitigating factor against fraud, if incorporated into the system. Just a thought.
A council is a high-resource endeavor which introduces other issues. My brain just works in a way to compartmentalized the assets from potentially malicious instigators. There are always tradeoffs.
The goal is to find that elusive 'optimal' balance.
There are only three kinds of people in today's world; People inventing scams, People running scams and People getting caught in scams ! ! !
Wait, I want to be in a fourth group: those constantly maneuvering to avoid scams. :)
Then you have three options ; )
move overseas , get filthy rich , or become a politician . . .
Plz let me know what you decide , I might join you if we can find a " two 4 one " deal : )
what a nice idea!
Great job @jesta. If @ned liked it have no worries it is approved. Your time was worth spending it.
It's things like this that make you a worthy witness! #25 now and climbing :)
Vote for @jesta as a witness! https://steemit.com/~witnesses
You have some great detail in this post. Good luck, and good luck to anyone else taking it on. A lot of work involved.
You have put in a lot of time and effort to come out with a well thought out proposal. Wishing you success in its implementation.
Very well thought out.
Thanks for heading this brainstorm!
From a developer standpoint, the project is quite complex.
It should be 100% secure and fraud-safe, embracing the Steemit model.
But it's always good to think big, implementing the whole thing in small steps.