>Selection of dog breeds for intelligence (German Shepherds, Jack Russell terriers, Poodles vs eg King Charles Cavaliers, Corgis) seems to have produced a large gap but it took a long time - dogs have been domesticated for ~5,000 years.

Dogs have been domesticated since the dawn of civilization (and probably before), but those specific breeds are relatively new. GSDs are just over a century old, Jack Russells slightly older, poodles several hundred years old and Corgis positively ancient at almost 1000. Note, too, that intelligence was not the sole quality they were bred for. The Russian Silver Fox is a great example of the massive changes that can occur in only a few generations if artificial selection for a single trait is performed.

That said, I agree with the gist of your argument: 15 pts of Iq per generation seems ludicrous.

The selection in Ashkenazi was not done in the same way proposed here - filtering took whole lifetimes. In this system, you compress the whole "live 50 years and have slightly more / fewer children" step into one procedure.

This one is equivalent to picking the best one of 50 naturally occurring children, and raising them alone, every generation.

If Michael Jordan he had 50 children with a similarly elite mother, most of them would regress to the mean - but the best one could conceivably be near his level. If he had only one kid, it's very likely that the kid would regress significantly.

Also, of course there are negatives to high IQ but most of the time, this selection method wouldn't be done for that level. Two people of IQ 100 would be able to reliably have children of IQ 115, and those kids would have happier, longer, healthier lives, with no increased risks. [see the scottish IQ study; iq at age 11 was linked to a lifetime of better outcomes]

That is the real benefit of this technology - to give people the option of gradually bringing out the best of what's already inside themselves. I don't want to be forced to give my kids a random selection of my genes - I want to exercise some control. And of course there could be problems - perhaps +IQ genes might lie next to other, undetected bad genes. But that's random, and we're already completely subject to it.

The Ashkenazi thing is bullshit. If you follow the references in the Wikipedia article looking for hard statistical evidence of higher IQ you end up with very thin sourcing from two iffy papers.

Wikipedia is not a reliable source, particularly on politically contentious matters such as this.

The Flynn effect in the U.S. already seems to be causing about 3 points of IQ gain per decade. If a generation is 25 years, that's 7.5 IQ points per generation. It doesn't seem unreasonable that if you explicitly selected for this trait you could double the rate of evolution.

http://en.wikipedia.org/wiki/Flynn_effect#Rise_in_IQ

> dogs have been domesticated for ~5,000 years

That's way too low. It's somewhere in the 19-32k year range: http://www.the-scientist.com/?articles.view/articleNo/38279/...

> <dogs domesticated> somewhere in the 19-32k year range

I have read numerous different accounts. It appears that the selective breeding of different breeds is actually much shorter even than 5,000 years. Most breeds are less than 1,000 years old.