Captcha Dao:
Goal is to incentivize captcha creators to design captchas that take 5 to 15 seconds to complete by humans. The captcha is proven bad if it can be solved often in under 5 seconds. The system has unique traits to prevent gaming of the system that is expected with open ended solutions to designing captcha software.
Captcha creators create captchas and must prove they are able to solve their own captcha in under 5 seconds 10 out of 10 times every 24 hours. This may sound counterintuitive, however, without this proof that the problem is solve-able, it is possible the captcha simply has a timer that waits exactly 5 seconds before allowing user input. Thus it is likely for captcha creators and miners to be the same group of people with the same expertise. Captcha creators also place a bond on their captcha where the size of this bond is used both as a reward for the miner that cracks it, and because the size of this bond will be incentivized to be correlated to quality, and thus enables purchasers to filter out spam or low quality captchas. Captcha creators may pull their software out of circulation. To prevent gaming, this requires 24 hours delay to allow miners to finish draining the bond if it was just solved to prevent creators from pulling before it can be drained.
Miners may attempt to prove a captcha is solve-able by bots by using a separate interface from the websites where the captchas are used with the general public. They pay a tiny bond which is returned to them double if they solve the captcha in under 5 seconds, and lose the bond if the time is over 5 seconds.
Captchas must be uniform by all allowing a randomly generated seed to deterministically create their captcha puzzle. Thus the same seed always generates the same puzzle. This enables process to be verified to be random on the blockchain.
Captcha Solver:------------------------------------------------------------------------------------
Benefits:
Gain access to the captcha purchasers website
Costs:
Time to solve captcha
Strategy:
Solve captcha as quickly as possible
Prediction:
Other than the captcha likely being far more creative than ones they are used to, they do not notice the difference compared to existing captchas.
Captcha Miner:------------------------------------------------------------------------------------
Benefits:
Payment for captchas that take less than 5 seconds to complete.
Costs:
Time to solve captcha
Bond to attempt captcha equal to the potential payment, returned if completed in under 5 seconds.
Strategy:
Solve captchas faster than 5 seconds in a way that is worth the time spent on solving captchas.
Prediction:
Bots will be created to target particular captchas, effectively mining.
These bots will likely use machine learning.
When a captcha is cracked, the miner will quickly drain the captcha creators bond until it hits zero, removing the captcha from circulation.
Captcha Creator:------------------------------------------------------------------------------------
Benefits:
Micropayment from captcha purchaser for captchas that take between 5 and 15 seconds to complete.
Higher visibility to captcha purchasers when larger bond is chosen.
If they choose to take their captcha out of circulation, they can recover their bond after 24 hour delay.
Costs:
Cost of bond to put a captcha in circulation. This is arbitrarily large, and may be added to at any time.
Partial loss of circulation bond due to covering cost of captcha payments to miners that solve in under 5 seconds.
Labor cost to create captcha software.
Strategy:
Create captchas that are completed by miners in 5-15 seconds more often than they are completed in under 5 seconds.
Choose a low price per use of captcha to attract captcha purchasers
Choose a high price per use of captcha to maximize revenue.
Choose a large bond to obtain higher visibility to captcha purchasers.
Choose a small bond to minimize losses when the captcha is cracked by a miner.
Prediction:
A market will form where a competition will be financially fueled to have ever increasing quality of captcha cracking and ever increasing quality of captcha.
Captcha Purchaser:
Benefits:
Filter out robots
Costs:
Cost per captcha for each captcha that takes between 5 and 15 seconds
Strategy:
Select captcha creators that offer a cheap cost and desirable average solving time.
DETAILS
Verification process of a successful crack:
Miner uses the mining interface.
Miner requests specific captcha to challenge it
Blockchain software creates a randomized seed for the miner to input into the captcha and makes it public with starting timestamp.
Miners computer reads the seed and runs the captcha software using this seed.
Miner solves the captcha in under 5 seconds and is given an output hash.
Miner publishes the output hash to system. This input and output hash are repeatable and can be verified to prove the captcha was solved in under the time provided.
System checks if the captcha was solved in under 5 seconds, (with no slack for internet communication time), and if so returns the miners tiny bond, and pays the miner an equal amount out of the bond for that captcha. If the captcha was not solved in under 5 seconds, the bond is transferred to increase the size of the captcha bond.
Captcha code is open source. There is no risk to being open source, the only code that the creator protects is the code they used to crack their own captcha.
Ok so we get a captcha creator to effectively say: I have created software that when it receives a seed it generates a puzzle that I expect will take 5 to 15 seconds to solve more often than it is to take under 5 seconds to solve. When the puzzle is solved correctly it outputs a hash that is deterministic based on the seed and timestamp. I have proven daily since I uploaded the captcha software that I am always able to solve this puzzle in under 5 seconds. Because I am confident it will take a long time for someone to automate a solution to this puzzle that can solve in under 5 seconds, I am placing a large bond online to signal my confidence and and increase visibility to captcha purchasers. Anyone who can prove they can do what I did and solve this puzzle more in under 5 seconds over 50% of the time will be awarded this bond and my puzzle will be removed. The continued existence of this bond is proof that only I am able to reliably solve this puzzle faster than 5 seconds. It stands to reason then, if someone places my puzzle on their website and it is completed in under 5 seconds, then either I am using my automated solution or someone else who has discovered an automated solution but is using it to bypass captcha instead of claiming the bond. Essentially, incentives prove that my captcha can only be solved in under 5 seconds by my software or other humans.
The marketplace this puzzle came from is better than existing captcha systems as it will only generate captchas that likely 2 or less people in the world have an automated solution for. This is likely better than current captcha systems. The competitive nature of this system will result in an arms race of machine learning and clever captcha creators.
I believe using coercion to defend ideas as one would for property is morally wrong. I however believe it is unethical (fraud) to falsely claim or imply one is the origin of an idea. If you gain value from ideas in this blog, I request you donate a portion of that value in an amount you feel is fair.