The cynic in me says that's exactly what's been done with humans. The school system in the United States was created, in large part, to train factory workers and be obedient citizens.
Much of what working dogs do is pre-programmed. It's instinctive. Very little (if any) training is required.
Why try to train things to be things they were not already designed to be? It's FAR more efficient to breed the traits and instinct you want into a breed.
Anyways, you can train poodles to be livestock guardians. http://www.vipoodle.org/HRP/VIP_herdingpoodles.html
[1]http://metro.co.uk/2017/04/23/britains-biggest-dog-is-7ft-an...
Of course I am. Do you think anyone can be a great basketball player?
Trying to pretend that all races are "the same" is blindness.
DNA is mechanistic in how it operates. Its power comes from our ignorance of the way it affects us. When we know how it works we're able to get in there and change our destiny. "Nature" and "nurture" aren't opposites, they're tightly integrated. To the extent they seem distinct it's because we are ignorant of how they relate in specific instances.
E.g. we might find that a particular gene is highly correlated with the early onset of kidney disease/failure. "Oh no!" says the person who finds out that they have a (say) 90% probability of kidney failure in before they're 60. What if, however, the entire reason that this gene led to kidney failure was because it decreased a person's desire to drink water? Low water consumption then led to stones leading to kidney disease. Given this information, the impact of the gene on kidney failure in an individual is entirely negated, because they can internally correct and just choose to drink more water.
This same line of reasoning can apply to intelligence: what if our genes code not for some mysterious magical "competence", but instead for a predisposition to find certain things interesting? And what if we're able to make the right things interesting in the right way? This enables us to significantly diminish the impact of those genes, too.
Similarly with dogs -- the impact of genes as an animal grows from being a baby to an adult is not on eventual outcome, it's on development. Some dogs are less trainable because they're less food motivated. What if, during the dog's development, we could slightly restrict access to food, and grow in the dog the desire for it? Now the dog is more trainable.
1. The Ethics Problem. Puppy mills are still around and still bad, and breeders in the past probably did some pretty terrible things to their dogs, but the human analog of the work performed by even the most humane dog breeders would still be a terrible travesty of the most basic human rights. Maybe I'm not creative enough or open-minded enough, but I can't conceive of a way to do this ethically.
2. The Nurture Option. A well-trained bulldog can be a much, much better dog than an untrained border collie. And one person can accomplish this training with a dog in a few years. Why spend 12,000 years breeding humans instead of putting a bit more effort into training the perfectly serviceable brains that we already have?
I tried to find some statistics to back up the latter point. The US spends about 2.7%, or 102 billion of the 3800 billion federal budget, on education. States average about 30% of their budgets, or (very roughly) 550 billion of the 1850 billion sum of state budgets, on education. A further complication is that much of education funding is done using local property taxes, which vary wildly by location, but are generally similar to the sum of federal and state funding. This is a total of about 1.2 trillion in education funding. Honestly, that compares pretty favorably to a 19 trillion GDP. And it ignores all the extremely important effort and money spent by parents to educate their children, and private spending on post-secondary education.