Well, objects fall, planets revolve routinely, creatures multiply, the moon rises and the sun sets. They think that much is perfect.
The trouble is that us humans tend to identify short-term order and think it means long-term order.
On cosmological scales, that means everything from an apparent constant night-sky, turning as if on the inside of a shell to looking at planets as if perfect points/spheres of light travelling upon circles (then extending that to perfect circles upon perfect circles upon yet more perfect circles, a practice that could eventually produce even a square path), to apparently eternal oval orbits, etc... Missing out a few stages of "paradigm shift", most of which occurred with the the intelligentsia and those whose positions required them to account for discrepancies discovered by themselves or pointed out by others.
And current scientific analysis has indicated that the planets in our solar system is indeed stable for a vastly long length of time (with currently observed effects, not counting some stray mass wandering through the system and perturbing/colliding with one or more bodies), so long in fact that the 'first' possible instance of instability we're going to encounter is after the Sun changing size as swallowing up the inner planets, anyway, and that probably leads to some effect to the rest but I don't have information on that at hand.
On shorter lengths of times, there are perceived patterns in the toss of a coin or the spin of the roulette wheel that gamblers swear by (e.g. "10 times on red? ...then it's more likely going to be black next time", by some. "...still red" for others. ...though 50/50 chance (ignoring for now the chance of green) in reality).
And the brilliance is that any perceived deviation from the 'perfect path' can always be rationalised. This can be done in both scientific and non-scientific ways, and looking back upon the history of science indicates many mis-steps along the way (the 'four element' theory of matter, dephlogisticated air, the aforementioned Ptolemeic concept of epicyclic orbits, the homunculus theory of human development, electrical charge and current, the very nature of vision and light, the list goes on) which would tend to indicate that even today's understanding of the universe is unlikely to be perfect and (to some extent at least) going to be invalidated at some point in the future.
And here's another thing about perfection: You know, there
will be a 'perfect predictor algorithm' for any apparent randomness, and there will also be many apparently perfect predictors which might appear to explain the randomness up to a point. For example, in trying to predict the output of a Pseudo-Random Number Generator, then obviously the perfect predictor would be an identical PRNG working with the same algorithm and seed applied in the same way, but ahead of time. But out of the almost limitless number of possibilities, there are going to be a small fraction (but still countless) other PRNG configurations of like type or otherwise (including processes including dice rolls and coin tosses, if the outputs can be mapped correctly) that match the first output of the target PRNG, a small fraction (but
still countless) of which will also match the second output... and so on.
Obviously the 'exact match' version of the test PRNG will continue ad infinitum to keep pace with the target PRNG, but there are always possible test PRNGs that could have copied (or presaged) the target one for any arbitrary number of cycles. And yet may
or may not have match with the PRNG on the next cycle.
Now, this could seem a million miles away from religion, but the efforts dedicated to performing a rite to guarantee a good crop in the coming season could easily be tied into how well that crop ends up doing. Similar to "wearing lucky underpants" for a sportsman (or, even more unlinkable to the outcome, a fan). Any differently-psuedo-random 'perfect' prediction that fails falls by the wayside (or is 'explained' or modified, or the inconsistencies are just plain forgotten about) but there are countless processes that remain in favour.
And, yes, burning a bushel of last years corn in the correct corner of the correct barn and observing the remains could provide a sufficiently correlated indicator as to next year's harvest, for obscure reasons beyond our understanding and in this case probably due to something beyond any current scientific understanding (whether its a deity taking note and causing the correlation or some aspect of solar activity being focussed by an obscure lump of rare ore built into the wall and affecting the combustion process in line with how the solar cycle will fluctuate in the months to come). And so a short-term order is identified and gone along with. Even though the long-term order is as yet unproven and merely assumed.
That is, of course, a cynical POV. I hasten to add that the theistic PRNG could
actually match the actual PRNG of the universe and
be perfect. If I believed in a non-deterministic universe, though, I wouldn't give much for its chances. (And, as I do, I'm think I'm as bound to believe/not believe what I do/do not anyway, regardless. And similarly for all you lot.