The Internet has enabled people to coactively create, share, rate and classify content on an unprecedented scale. Examples of coactive systems include open source software development, wiki communities, social news aggregators, and many others. This paper discusses regularities in online coactive systems of thousands or millions of participants. First, user participation levels are shown to follow a heavy-tail power-law distribution over their entire range, so that a small number very active users make the vast majority of contributions. The power law arises from a simple rule where the probability a person stops contributing varies inversely with the number of contributions. The power law exponent is moreover demonstrated to discriminate between systems according to the effort required to contribute. Next, the level of activity per topic is shown to follow a heavy-tailed distribution, this time lognormal, generated by a simple stochastic popularity reinforcement mechanism. The vast majority of activity thus occurs among a small number of very popular topics. The trends are demonstrated to hold for four large independent online communities with different scopes and purposes.