Users can join, share a tag into the group (sharing system)
Each entry of this tag is vetted based on ML to determine eligibility
The entry's n_tokens is used to determine payout to user, from group.owner (researcher)
How to handle Behaviors cost?
Details
HUGE NOTE! This is HIPPA, sensitive, privacy, all the stuff. Never in my wildest wildest dreams would I ever sell this data from your profile without your consent. And no privacy-policy fine-print you agree to either. This feature would be explicitly between a user and a research group, where the user opts in to the group, sees how their data is used, agree accept agree yes. Data is encrypted / anonymized / scrubbed however it needs to be.
The groups feature coming soon. There would be two types of groups:
No data-selling, group owners are more like consultants, offering services (video sessions, personalized journal responses, etc). In this case, the members pay for the services (or not, group owners can make things paid or free).
Group owner is a researcher, and the members offer their entries for pay. Originally I thought to make this a blockchain marketplace, and the payout is something of a coin - but I don't think this is necessary, let's just use Plaid for the researcher's account for the payout.
There's a lot of value in sharing Behaviors to researchers. Eg, using the Behaviors feature to track medications and their impact on various other fields (mood, sleep, etc). Tracking medication dose changes, interactions between medications, their impact on detected sentiment in your entries - all via causal inference #23. I don't yet know how this would be monetized between researcher & user, I'd love some thoughts.
More simple is the Entry exchange. The researcher sets a price-per-token. When a user submits an entry, if it has a tag which is shared into this group (which is how sharing with groups works generally), then the following happens:
The entry is vetted, to ensure it's not a bot:
Cluster the entry into some global cluster system to make sure it's not an outlier; eg not a Wikipedia copy/paste, definitely a journal entry. Ensure it's also no more than x-distance from the user's other entries.
Use LLM to confirm if it looks like a real entry
The entry is computed cost-per-token, money sent from researcher's pool to user
Possibly the researcher has some follow-up questions / surveys, or interactions within the group via chat, video sessions, etc (additional payment to user or part of the whole package).
TL;DR
Details
HUGE NOTE! This is HIPPA, sensitive, privacy, all the stuff. Never in my wildest wildest dreams would I ever sell this data from your profile without your consent. And no privacy-policy fine-print you agree to either. This feature would be explicitly between a user and a research group, where the user opts in to the group, sees how their data is used, agree accept agree yes. Data is encrypted / anonymized / scrubbed however it needs to be.
The groups feature coming soon. There would be two types of groups:
There's a lot of value in sharing Behaviors to researchers. Eg, using the Behaviors feature to track medications and their impact on various other fields (mood, sleep, etc). Tracking medication dose changes, interactions between medications, their impact on detected sentiment in your entries - all via causal inference #23. I don't yet know how this would be monetized between researcher & user, I'd love some thoughts.
More simple is the Entry exchange. The researcher sets a price-per-token. When a user submits an entry, if it has a tag which is shared into this group (which is how sharing with groups works generally), then the following happens: