This is a stunning visual - thank you. I see a multi-dimensional space of possible organism "configurations". Each dimension is a set of related alleles, spaced by the amount of energy it would take to shift from one to the other, i.e. the likelihood of switching that allele independently. Angles between dimensions indicate correlation, or more fundamentally the cost of holding similarly positioned alleles simultaneously. Each organism is a point in that space with a species being a cloud.
When one shines a "light", i.e. an antibiotic, on the space it lights up with warmer colours signifying configurations which take less energy for the organism to survive as (are more favourable) and cooler colours denoting deadlier configurations.
Thus, antibiotics become a game of chess. Humans try to box the organism into coolly coloured regions. That way, the amount of energy it would take to "tunnel" through the wall would be high. The organism, on the other hand, is trying to find warm "tunnels" by which to escape regions which are being boxed in. Advantage to the organism is it can build new dimensions.
However, it's one that doesn't need an agency to implement it. It just happens. A lot. Especially in the case of bacteria. You have a slow, crappy algorithm, but it's constantly running relentlessly in parallel in many trillions of instances. It doesn't need human attention. It doesn't need investment. It doesn't need society to be aware of its importance.
On the other hand, progress in antibiotics needs all of the above. If one of those things flags a bit, bacteria gain the upper hand. This is the problem.
It took evolution billions of years to make machines that could make machines that run about 700 miles per hours.
You come off as being dismissive of the power and beauty of evolution, although I don't think that's your intent. Evolution assembled the most complex system known to exist, starting from a handful of molecules, without any "thought" or "reasoning." The entire concept/illusion of life and consciousness was fabricated out of simple natural elements and physical processes. Amazing. Sure, once all of that has been put in place, it's trivial to expand upon. I'm sure you agree, I just like to take any opportunity to point this out. :)
If I come across as cynical about this, it's a response to the way evolution as a process has been fetishized by people who don't understand it. People look at the application of evolutionary algorithms to solving difficult problems and conclude that evolution is smarter than humans. But evolution's power really comes from its dumbness. It works without understanding the problem, and so it can be used to solve problems we don't understand very well. But when we learn how to model a problem intelligently, we're generally able to crush the results of evolution.
It's a recurring pattern. SAT, the NP-complete problem, is at first glance a ridiculously simple task, yet solving it means being able to solve all other problems. When designing a language or a machine, it is very very simple to accidentally make it Turing compatible, the barrier is just so low, despite the reward being.. everything.
There may have been a lot more complexity early on than that statement assumes. Static systems, after all, pretty much by definition can't evolve (and this is why a reductionistic assessment of evolution as biochemistry doesn't work).
The beauty behind evolution is the same beauty behind Feynman integrals in quantum physics. You take the integral over all possible paths and something is likely to work. Many paths will end or cancel eachother out.
It's like open source software.
What's really huge and what no is really a limit and perception of our brains, no?
For a evolutionary process that has started at the earliest stars after the Big Bang. Those species are currently so powerful, that we are very likely to them the same as bacteria are to us. Might as well perceive that to be a very reasonably efficient process.
Some problems are just very difficult to evolve around, and it's hard to predict what they'll be even if you have a complete working knowledge of an organism's genome and biological workings. If the problem space doesn't contain a solution that can be climbed toward without sacrificing fitness in the short term, it is unlikely that it will ever be found by evolution.
Evolution is actually a really crappy optimization algorithm. The reason it has worked so well is not that it is so effective, but because it's the only game in town. It took evolution billions of years to make machines that can run 60 miles per hour. It took humans a few thousand years to create machines that run about 12 times that fast.
Bacteria work on a faster time scale than animals in terms of evolution, but five orders of magnitude? Somehow, I think we'll win in the long run.