Recognition works when it proves somebody read the contribution: name the specific thing, say what it changed, and do it within a day or two. Points and leaderboards measure volume instead, which is the one thing a small community does not need more of. Recognise the work in public and the person in private.
Flow diagram of 5 steps: Name the contribution, not the contributor, Say what changed, or say plainly that nothing did, Do it within a day or two, while the work is still live, Recognise the work in public and the person in private, Offer responsibility rather than status, where it is wanted.
The word recognition covers two different things that behave in opposite ways. One is a reward: something handed over in exchange for work, which quietly establishes an exchange rate and invites the recipient to ask whether the next piece of work is worth the price. The other is a signal that the work was read by somebody who understood it. Communities almost always build the first, because it is the one you can build — a counter, a badge, a rank, all of which run without anybody remembering to do anything — and almost always need the second, which cannot be automated at all, because the entire content of the signal is that a person paid attention.
The distinction is testable. Ask what a contributor learns from the recognition they receive. From a points total they learn how much they have posted, which they already knew. From a sentence naming the specific thing they said and what it changed, they learn three things for the same thirty seconds of somebody else’s effort: that a person read it, that it had an effect, and roughly what kind of contribution counts for something here. The third is the part that gets underrated. In a community with no training and no manager, what gets noticed out loud is the only norm-setting mechanism there is, and people do more of whatever they saw someone else get noticed for.
What each form of recognition actually tells the contributor
| Form | What it signals | What it fails to signal |
|---|---|---|
| A points or contribution counter | How much you have posted, which you already knew | Whether any of it was useful, or whether a person read it |
| A badge or role awarded at a threshold | That you crossed a line the software was watching | Anything about the specific work — the same badge is worn by people who did unlike things |
| A public thank-you with no detail | That somebody wanted to appear appreciative | Which contribution is being thanked, so it reads as politeness rather than attention |
| A sentence naming what you said and what it changed | That a person read it, understood it and acted on it | Nothing much — this is the form that works, and it does not scale |
| A decision publicly attributed to your argument | That your participation altered an outcome | Nothing, though it is only available when there was a real decision to attribute |
Ranking systems are attractive for good reasons: they are cheap to build, they need no upkeep, and they produce a number that goes up. The objection is not that they get gamed, although they do. It is that the thing they measure accurately — volume — is close to the opposite of what a small community needs more of.
Take two members over a month. One posts forty short replies, most of them agreement, and sits at the top of any counter you care to build. The other reads three submissions properly and writes one careful assessment of each, telling two authors their scope is too broad and sparing the community a commission that would not have held together. The second member is worth more and scores about an eighth as high. Nothing has malfunctioned; the counter measures exactly what it claims to. Posts are easy to count and judgement is not, and a system that counts only the countable ends up announcing that quantity is what is wanted here.
After a few months, the top positions on a cumulative board are held by whoever joined earliest and posts most, and the gap is arithmetically unbridgeable for someone who arrived last week. That is not a motivation problem to be patched with a monthly reset — it is simply what a running total does. The effect on newcomers is the reverse of the intention: the board tells them precisely how far behind they are and how long it would take to matter. Rankings also convert a set of collaborators into competitors for a scarce position, which is an odd thing to engineer in a community whose entire output is people helping each other with their work.
When a number confers standing, some people optimise for the number, and it is worth being clear that this is usually not cynicism. Publishing a board is a statement about what the community values, and posting more is the honest reading of that statement. What follows is predictable: shorter replies, more of them, threads padded with agreement, and a slow drift towards whatever the metric happens to reward. The real damage is not the few obvious cases. It is that the metric stops describing who is actually contributing, which was the only reason to have it.
The practical version of this is unglamorous and costs a couple of minutes a day. There is no artefact at the end of it, which is why it usually loses out to building a badge system that takes a week and then runs itself.
Thanking somebody for being a valued member says nothing about anything they did, and the recipient can tell. Naming the thing — the point about audio quality, the objection to the running order, the review that caught a wrong date — is the whole content of the signal. It also takes no longer to write, because the specific version is a description of something you have just read, whereas the generic version has to be invented from nothing.
The strongest recognition available in a community is a decision traced back to the argument that caused it: the format changed because of this objection. Where an argument changed nothing, saying so beats both silence and a vague thank-you. It tells the person their contribution was read and rejected, which is information they can use, whereas silence tells them nothing and is generally read as the worse of the two possibilities.
Recognition decays fast. A note about something written yesterday is a response; the same note three weeks later reads as an audit, or as somebody working through a backlog of appreciation, and it tends to land as faintly awkward for both parties. Given a choice between a one-line acknowledgement today and a considered paragraph next week, the one-liner is worth more.
These want different venues, and collapsing them is what turns recognition into a popularity contest. Quoting somebody’s argument in a public thread and saying what it changed is about the work, is useful to everyone reading, and transmits the norm. Telling somebody they have been holding a lot of this together belongs in a private message, where it creates no visible hierarchy and does not put the recipient in the position of having been publicly designated as important.
For some contributors the most convincing recognition is being asked to do something that matters: review the next batch of submissions, set the running order, write the rule that was missing. That is a promotion in trust rather than a decorative title. It has to be optional and revocable without embarrassment, though, because handing someone unpaid responsibility they never asked for is a way of losing them that looks like an honour on the way past.
One category of contribution is under-recognised in every community, and not for want of goodwill. It is that the work leaves no artefact behind, and recognition tends to follow artefacts.
When somebody answers a hostile comment carefully and the thread quietly ends, what remains afterwards is a short thread. Compare the member who wrote the long, sharp, quotable reply: that is visible for as long as the thread exists and will be remembered as a contribution. The person who prevented an argument has produced an absence, and absences are hard to thank. This is worth compensating for on purpose, because it is precisely the work a community most needs and least rewards, and the people who do it are the least likely to point at it themselves.
Someone who reads every submission and then votes carefully leaves a trace indistinguishable from someone who logs in and clicks at random. A community that recognises only writing has told the careful reader they are not really participating, which is false: the counts mean something only because some of the people casting votes read the material first. No counter can see this distinction, which is a further reason not to let a counter serve as the community’s account of who matters.
Fixing broken links, correcting a wrong date, keeping a list current, answering the same beginner question for the ninth time — none of this is noticed being done and all of it is noticed being undone. It also has no natural moment of completion at which somebody could be thanked, unlike a submission or a review. The only dependable fix is for a person to look, deliberately, for the contributions that generated no post. That makes recognition a task somebody owns rather than a feature of the software.
Recognition cannot compensate for a community where nothing is decided. If contributions do not visibly change an outcome — if the votes are advisory in a way nobody admits, or the same person picks the result regardless — then attributive recognition has nothing true to attribute, and warmer thanks make the gap more conspicuous rather than less. That problem is upstream and structural, and the fix is to put a genuinely open decision in front of members. Recognition is a way of showing people the effect they had; it is not a substitute for their having one.
It also inflates. If every post is acknowledged, the acknowledgement stops carrying information, and the community has spent real effort to arrive back where it started, now carrying an expectation it has to keep meeting. There is a harder limit worth stating plainly as well. Where the work has become substantial and continuous — somebody effectively running a strand of programming, or moderating every day — recognition is the wrong instrument entirely. That is a job, and offering appreciation in place of pay or a formal role is a way of extracting labour that both sides can feel and neither wants to name. The honest options at that point are to reduce the load, share it out, or pay for it.
On the evidence behind this page: Kind Channel is new, its community is small, and no programme has aired yet, so nothing above is drawn from our own participation data. We do not have enough of it for a figure about returning contributors to mean anything, and any percentage we published would be invented. The comparison of two members earlier is an illustration, not a measurement. What this page offers is the ordinary observable behaviour of forums, editorial groups, open-source projects and volunteer organisations, written as a practitioner’s account — treat it as that, and be sceptical of any new platform offering you retention statistics it cannot yet have collected.
They reliably increase volume and do not reliably increase usefulness, which matters whenever the two diverge. A counter can only measure what is countable, so it rewards whoever posts most rather than whoever helped most, and it announces to everybody that quantity is what this community values. Cumulative boards also have a top that early joiners hold permanently, which mainly tells newcomers how far behind they are.
A reward is handed over in exchange for work, which sets up an exchange rate and invites the recipient to weigh whether the next piece is worth the price. Recognition is evidence that a person read the work and understood it. The first can be automated with badges and counters; the second cannot, because the entire content of the signal is that somebody paid attention.
Split it by subject. Recognition of the work — quoting an argument and saying what it changed — belongs in public, where it is useful to everyone reading and transmits what this community values. Recognition of the person, such as telling somebody they have been holding a lot of this together, belongs in a private message, where it creates no visible hierarchy and does not publicly designate anyone as important.
By looking for them deliberately, because they will not surface on their own. De-escalating an argument, reading submissions carefully before voting, fixing broken links and answering the same beginner question repeatedly all produce no artefact, and recognition normally follows artefacts. That makes it a task a person has to own rather than a feature of the software: somebody must go looking for the contributions that generated no post.