4 ms·

One thing to keep in mind is that evolution does not produce global optimums - only local ones.

by Fomite 5y ago

dmichulke 5y ago
That's kind of a hard thing to measure against, can you tell me of an example in real life where you know something is the global optimum?
- ben_w 5y ago
  That’s basically why it does local optima rather than global. As for an example where you know the global optima: real life hill climbing rather than the blindfolded version used to explain the algorithm of the same name?
- Fomite 5y ago
  You don't need to know something is a global optimum - you just need to know there's a slightly better at-least-local optimum with a decrease in fitness between that and where one is now.
hypertele-Xii 5y ago
While the individual interactions between organisms are indeed local, sexual evolution precisely attempts to work around this limitation with selection pressure for diversity. That is to say, sexual evolution increasingly approximates a globally optimal process, as an aggregate system of autonomous agents. The Internet is also not actually a global network. It's just lots of local connections between two computers. Yet the effect approaches a global network.
- yunohn 5y ago
  You use a lot of jargon, but I’m not sure where you ended up. It seems obvious to me (from logic and observation) that most animals don’t have global connectivity (despite your comparison to internet meshes), and obviously evolve differently across the world.
  - hypertele-Xii 5y ago
    There is genetic exchange between different animals from across the world, which is a mechanism that has evolved presumably specifically because it strengthens global resolve. It's entirely non-obvious but genetic material propagates through many pathways, including but not limited to sexual reproduction. Virii, for example, transfer genes between species. Bacteria have plasmids, rings of genes they can literally just give to other bacteria.

One thing to keep in mind is that evolution does not produce global optimums - only local ones.

by Fomite 5y ago

One thing to keep in mind is that evolution does not produce global optimums - only local ones.

That's kind of a hard thing to measure against, can you tell me of an example in real life where you know something is the global optimum?

ben_w 5y ago
That’s basically why it does local optima rather than global. As for an example where you know the global optima: real life hill climbing rather than the blindfolded version used to explain the algorithm of the same name?
Fomite 5y ago
You don't need to know something is a global optimum - you just need to know there's a slightly better at-least-local optimum with a decrease in fitness between that and where one is now.

hypertele-Xii 5y ago

While the individual interactions between organisms are indeed local, sexual evolution precisely attempts to work around this limitation with selection pressure for diversity. That is to say, sexual evolution increasingly approximates a globally optimal process, as an aggregate system of autonomous agents. The Internet is also not actually a global network. It's just lots of local connections between two computers. Yet the effect approaches a global network.

yunohn 5y ago
You use a lot of jargon, but I’m not sure where you ended up. It seems obvious to me (from logic and observation) that most animals don’t have global connectivity (despite your comparison to internet meshes), and obviously evolve differently across the world.
- hypertele-Xii 5y ago
  There is genetic exchange between different animals from across the world, which is a mechanism that has evolved presumably specifically because it strengthens global resolve. It's entirely non-obvious but genetic material propagates through many pathways, including but not limited to sexual reproduction. Virii, for example, transfer genes between species. Bacteria have plasmids, rings of genes they can literally just give to other bacteria.