Community members decide, by voting. Every submitted idea first receives written AI editorial feedback, which never blocks anything and casts no vote. Ideas then go to the community, and the best-supported are scheduled as programming. Staff handle whether something can practically be produced, not whether it deserves to be.
Flow diagram of 5 steps: Submit the idea, Read the editorial notes, Revise and resubmit, as often as you want, The community votes, Feasibility is assessed.
It is worth being precise about this, because "AI-assisted curation" is easy to misread as an algorithm choosing what people are allowed to see.
What each participant controls
| Input | What it does | What it cannot do |
|---|---|---|
| The AI editor | Returns written notes on five criteria | Reject an idea, or vote on it |
| Community members | Vote and comment on submitted ideas | Skip the editorial feedback step |
| Foundation staff | Assess whether an idea can be produced | Override the community vote on merit |
The order matters as much as the roles, because each stage deliberately limits what the next one can do.
A submission is a written proposal rather than a finished piece. It has to describe what would be made clearly enough that somebody voting can picture it — a lower bar than a script, and a considerably higher one than a title.
Written feedback comes back across five criteria with the reasoning shown rather than hidden behind a number. Nothing is blocked at this stage and no vote has been cast. The notes exist so the proposal can improve before anybody judges it publicly.
There is no limit on revisions and no penalty for using them. This is the cheapest point in the entire process to change your mind about what the idea actually is, and it is the stage most people skip.
Members vote on submitted ideas, and this is the stage that decides merit — the only stage that does. Kind Channel has not published a vote threshold or a fixed voting window, and this page is not going to invent one.
For a well-supported idea, staff work out whether it can actually be produced: whether the people in it consent, whether it carries legal risk, whether the resources exist. This stage can stop an idea, but only on practical grounds.
An idea with community support behind it that clears feasibility becomes scheduled programming. This is the stage where the process is least proven, for the straightforward reason that it has not yet happened.
Everything above describes how the decision is meant to work, and it is worth saying plainly that this is a commitment rather than a track record. Kind Channel is new, the community is small, and no broadcast has aired. Nobody can point you at a produced programme and show you the vote that produced it.
That should affect how much weight the governance claim carries. A rule that has never been tested under pressure has not yet had to hold — and the pressure, when it arrives, will look like a popular idea that staff privately dislike, a vote that lands somewhere awkward, or a feasibility call that is indistinguishable from a veto unless the reasoning is published. The design is sound and unproven, and the first genuinely contested decision will tell you more about it than this page can.
Until then, the checkable parts are the ones worth checking rather than taking on trust: the editorial reasoning is visible for every idea assessed, vote counts are shown as they were cast, and the one stage where staff can intervene is stated here rather than buried.
A language model is reading text, not verifying it. It cannot know whether your account of your own life is accurate, whether a community wants what you are proposing, or whether you can deliver it. Giving something with those limits the power to reject submissions would filter out exactly the unusual ideas the platform exists to find.
So it advises and shows its reasoning, and the reasoning is visible in the Director Studio for every idea assessed. You can disagree with it in public, which is the point of showing it.
Once the community has chosen, someone still has to work out whether the thing can be made — whether the people involved consent, whether it raises legal risk, whether it is possible with the resources available. That is a feasibility judgement, and it is the only stage where staff can stop something.
What staff cannot do is set aside a well-supported idea because they personally find it uninteresting. If that ever happened, the governance claim on the rest of this site would be false.
No. It never blocks a submission and never casts a vote. It returns written editorial notes across clarity, originality, social impact, engagement and feasibility, and you can revise and resubmit as many times as you like.
Yes, a free one. Voting is attributed so that results mean something, and browsing ideas requires no account at all.
It stays in the pool rather than being discarded. Feasibility problems are often about timing, access or consent, and those change. The vote is not wasted.
No. Counts shown are the counts cast. Kind Channel is new, so these numbers are currently small, and inflating them would be a straightforward misrepresentation of how much support an idea has.