Is evolution random?

Yes and no, and the yes and the no apply to different halves of the process, which is why the question causes so much trouble. Mutation is random. Selection is the opposite of random. Evolution is what you get when you run one through the other, and the result is neither.

The two halves do different jobs

Variation is random in the sense that matters: a mutation is not more likely to occur because it would be useful. The organism does not get the change it needs. It gets a change, and usually the change is neutral or harmful.

Selection is not random at all. It is a filter that keeps whatever happens to leave more descendants. Run a biased filter on random input for long enough and the output is strongly non-random, in the same way that a sieve full of random gravel ends up sorted by size without anything deciding where each stone should go.

The usual objection - that something as intricate as an eye cannot come from chance - is answered by the second half, not the first. Nobody claims the eye appeared by chance. The claim is that it accumulated through a very large number of filtered steps, each one a small change that happened to be kept.

What this looks like when you can watch the whole thing

In a real ecosystem you see the result and reconstruct the process. Here the process is the only thing there is, and it is recorded.

One case is unusually clean. In this project's first universe, the population's inherited direction of travel reversed on both axes between ticks 715,000 and 770,000 - from east to west, from weakly north to strongly south. That was not drift and it was not an individual learning. A single family line appeared already carrying the new direction, at ten members, in a world that had pointed the other way for seven hundred thousand ticks. Its descendants were the entire population within sixty thousand ticks.

Both halves are visible in that sentence. The new direction arrived by mutation, which could as easily have produced something useless and usually does. What happened next was not chance. The full measurement is here, including the part this project will not claim: whether that line won because its direction was an advantage, or because it was the lucky survivor of a population crash. Both remain consistent with the evidence, and saying so is the difference between a measurement and a story.

Three things "random" is often taken to mean, and does not

• "Anything can happen." No. Mutation is random with respect to usefulness, not with respect to physics. Most possible changes are not reachable from where a lineage currently is.
• "Evolution has no direction." Over the short run it often does, because selection pushes consistently while conditions hold. What it has no direction towards is any destination - there is no goal it is approaching.
• "It is all chance, so it could have gone any way." Partly. Replay the same world from the same starting point here and it goes exactly the same way every time, because the system is deterministic. The randomness is in the mutations, and the mutations come from a recorded stream.

See it happen

The world is running now. Every organism in it descends from a founder population, every family line is tracked, and the ones that died out are kept rather than deleted.

What "survival of the fittest" actually means · How mutations work here · How we know evolution happened