It assumes that biological chemistry is the same as engineering.

You can hammer a glob of ore forever with a bolt of lightning, and never get a single gear. But hit a pile of simple chemicals with some energy, and see what happens.

Don't confuse statistics with chemistry. If I took a very large, flat steel box, and laid pennies all across the bottom of it, how long would I have to shake the box before all the pennies came up heads?

Now, stick a magnet to the "tails" side of each coin, and ask the same question.

If you'll pardon the anthropomorphization, atoms WANT to make structure. You can't look at a display of crystals, or a snowflake, without seeing this. How long should it take for simple random assembly of water molecules to make something that looks like a snowflake? Yet billions are made every day.

Another thought experiment: Put a million monkeys into a room with typewriters. How long to complete a Shakespeare play?

Now put those monkeys in front of computer terminals, and run their output through an electronic program. This program recognizes english words, sentence structures, and phrases, and if real words and sentences are created, this is laid down, and the additional input applied to that information. Now how long are we talking? Still a long time, but nowhere near the time required by the original concept.

Again, you can hardly KEEP atoms from forming more complex molecules. The "watchmaker" metaphor falls flat.