Volunteer groups run on something that looks fragile from the outside. Nobody can be told what to do, nobody has to show up, and the people doing the most work are often the ones least able to say no. Community gardens, mutual aid networks, sports clubs, open source software projects and neighbourhood associations all share this basic shape, and they tend to succeed and fail for the same handful of reasons.
Those reasons have been studied more carefully than most people realise. Open source software projects in particular leave an unusually complete record, because almost everything happens in public and can be counted afterwards. Researchers have used that record to answer questions that are hard to study anywhere else. How long does a newcomer wait before someone replies to them, and how much does that wait matter? What proportion of first-time volunteers ever come back? What happens to a group when the person holding it together quietly stops showing up?
The answers travel well beyond software, because the underlying problem is the same everywhere. A group that depends on goodwill has to make it easy for people to give that goodwill, and most of the ways groups fail at this are unintentional.
These guides collect what the evidence shows.
Start here
Communities that describe themselves as welcoming often are, for most people. What makes a community feel welcoming looks at what happens when the same survey responses are separated out by who gave them, and why an encouraging average can conceal a group that several kinds of people are quietly leaving.
Every group has a way of making decisions, and most have never written it down. Who actually makes the decisions describes the four arrangements almost all volunteer groups use, and a few questions that quickly reveal how a group really settles things.
Most people who consider volunteering never complete a first contribution. Why new volunteers quit before they start follows the sequence from first interest to first finished task, and identifies the points where people reliably drop out. One recent study suggests those points have been getting worse.
Groups tend to hand leadership to whoever does the most visible work, then wonder why it goes badly. Leading when nobody is in charge covers what the job actually involves when you cannot instruct anyone, and how to make it survive the person currently doing it.
Long-running groups argue about whether things were better before. Why communities change as they grow separates the parts of that argument that hold up from the parts that mistake one problem for another entirely.
Who attends an event is mostly settled before the programme exists. Running events people can actually attend works through the decisions that determine it, nearly all of which are made months in advance and come down to money.
A note on the numbers
Figures here come from the original research wherever it exists, including community surveys by the Linux Foundation and Tidelift, published statistics from long-running mentorship programmes, and peer-reviewed studies of how volunteer contributors behave. Where a number is disputed or the sample is small, that is said plainly.
Two limits are worth knowing up front. The most widely quoted figures on who participates are getting old, and nothing comparable has replaced them. And research tends to study large, well-organised groups, because those are the ones that keep records. What holds for a national organisation with paid staff frequently does not hold for four people running something in their spare time.