Actually, it can work rather simply: pay people who have certain extreme, genetically-dominant conditions to let you sterilize them.
Here is an article written by a man with osteogenesis imperfecta about how he supports eugenics to try to prevent people with his condition from breeding.
(Yes, it's the Daily Stormer, but it's not written by a Nazi, and it's not a racist or terribly political article.)
Again, that
doesn't work. It prevents one person from having kids; it doesn't eliminate the condition from the gene pool.
What matters, ultimately, is the level of
selection pressure that the program can produce. As I noted, these are negligable over human timeframes, and maintaining the program for multiple generations is pretty much impossible from a pragmatic perspective. On top of that, it'd be pretty much impossible to even implement the program over the entire gene pool, so the trait in question would never be completely eliminated
anyway and would start coming back as soon as the program ended -- a few generations, and it's like you'd never tried your program at all.
Of course, that's assuming that heredity is the only factor... and it
pretty much never is.
Y'see, a whole lot of genetic conditions, especially the autosomal dominant ones, are caused by something called "
de novo mutations". This means that the person who has the genetic condition has it... but
neither parent did. Estimates on the number of de novo mutations each of us has vary, but they typically range from 50-100 per person on average.
I usually use
Huntington's Disease for this part of the explanation, but your article uses OI -- and the effect's even more extreme (if also more complicated) there.
OI, you see, is actually classed as eight different conditions, and is based on mutations to six different genes. It's thus more a conglomeration of different genetic diseases than a single condition... but your argument only talks about conditions which are dominant, so that throws out Types VI-VIII, which are recessive. I also don't have data offhand on the rate at which de novo mutations cause Type V, so let's focus on Types I-IV. Conveniently, they're all caused by mutations to two genes --
COL1A1 and
COL1A2 -- so they can be (sorta, kinda) lumped together as a single condition.
(Type V is caused by defects in
IFITM5, which is located on a different chromosome entirely. Like I said, I don't know the de novo mutation rate there and can't be arsed to look it up for something like this when it's almost entirely beside the point.)
Roughly 60% of cases of Type I and Type IV OI are caused by de novo mutations. That means that 6 out of 10 people with those conditions wouldn't be effected -- at all -- by your hypothetical program. At the
absolte best, you'd reduce the incidence of Type I and Type IV OI to 60% of what it currently is.
This suggests, by the way, that there's
already pretty damned severe selection pressure against Type I and Type IV OI. The mutated genes are still in the gene pool
because it keeps being eliminated and reintroduced. No eugenics program can do the slightest damned thing about them.
And Types II and III? Pretty much a
ll of those cases are caused by new mutations --
nobody with them reproduces anyway for the very simple reason that people with them don't survive to do so.
In other words, eugenics programs like that
don't work.
Now, I mentioned Huntington's? The de novo mutation rate there is 10%. Most genetic conditions -- especially the autosomal dominant ones -- have similar issues. Programs like you're talking about
simply don't work... and that's even before I get into things like side effects.