This is definitely an idea worthy enough to perfect and realize. However, looking at every response and deciding which contribution stems from which previous contribution and assigning each contribution an unbiased appropriate weightage of the final solution is a moderation-heavy and, more importantly, subjective task. If we make it more objective, not only will it be more unbiased, but an algorithm may be able to assign appropriate weightage, reducing the burden on moderators.
There should be some amount kept for the winning contribution. As an example, 30% of the bounty goes straight towards the winning contributor. The remaining 70% is distributed among the other contributors based on their help in reaching the final solution. How do we objectively calculate each contributor's contribution? I initially thought of using words as a "contribution unit". Every solution will have words and when you read the previous contribution, you use the same words and add your part to build on it. Therefore, words could be used to distribute the bounty appropriately.
For example, if I am one of the initial creative contributors, the unusual words (after removing pronouns, articles, conjunctions, etc., and also the words used in the challenge) in my contribution will be listed. When further contributors use any of these words, it gets recorded. Let us say that I used the word "monoclonal antibodies" in my contribution. If in the final solution that wins, there are 100 words and "monoclonal antibodies" appears 5 times, it is assigned 5% of the bounty. Now, since I am the first user of the word "monoclonal antibodies", I get 50% of the 5% bounty. The rest of the 50% is distributed equally among the chain of contributors who have used "monoclonal antibodies". This leads to a fair distribution of the bounty to every contributor.
There are some loopholes here. For example, someone who has no idea how to solve the challenge simply tries to use the words used by previous ideators and gets at least some part of the bounty. How do we identify such freeloaders? I thought of eliminating every contribution from the bounty that has a threshold of previously used words. For example, if your contribution has 70% or more words that were used by previous contributions, your contribution does not receive any part of the bounty. Will this work?