[CuriousSkeptic]

Statistics.

Imagine any given configuration of the universe. Of all possible changes to a new configuration (think Levenshtein distance) from that configuration there’s a large set of “equivalent” (in the aggregate) configurations, let’s call those the future. There is a smaller set of configurations which we would ascribe some special character too, such as being somewhat orderly let’s call those the past. the set of possible configurations encoding a broken pot are larger then the set encodings an unbroken one. But the state brokenness of the pot is entirely a subjective things, without that interpretation of the encoding they are equally random

Now if you are a “process”, a kind of pattern that can be identified as the same entity in several of those configurations, it would seem necessary for that pattern to follow some rule in which most of those “future” configurations would retain its unique characteristic.

So if you throw a dart into this mess of configurations and hit such an entity, selecting any direction to move from that point at random would most likely give you a configuration with higher entropy, but in which this particular pattern would be retained.

Some patterns would have a higher chance than other of being traceable along those changes, perhaps they encode a kind of anticipation of likely futures yielding a rules set which a higher likelihood of existing from one configuration to the next. Let’s call this patterns “living”