This is one of my favorite stories. I remember watching the documentary they did on this.
If I say "on average American men are 5 feet 9 inches" and you say "and yet we have Robert Pershing Wadlow who is 8 feet 11 inches"
That just doesn't make the case you want it to make.
We also have the famous Williams sisters of tennis fame who's dad was their coach too.
However, let's focus back on the original articles claim about the "science behind making your child smarter". Specifically we need to zero in on the "make" portion of the claim. That there is something that you as a parent can do to engineer the outcome of having your children be smarter than they otherwise would.
Let's evaluate the case of the Polgar sisters as it relates to that claim. Well, if look at the Williams sisters and the Polgar sisters together we could make the case that it might , perhaps, be possible to engineer the outcome of your children becoming world-class at something. In that case I would expect to see a pattern within categories and not just between, and I would also expect more data points. Perhaps more data points are out there and I would be interested to see as many as possible. The data we really need which we don't have is the data which would shed light on the survivorship bias - "how many tennis coach parents or chess grandmaster parents tried to engineer the same outcomes and failed?". For all the tennis parents out there and for all the effort those parents and their kids put in, their kids performance probably falls on some normal distribution of achievement in the sport. You can generalize that to whatever profession you'd like and for the claim to be true that there was something you could do to engineer the outcome you would want to see a high success rate or at least for the cases that we can find not for them to be statistical outliers. My guess is that they are, but I'm open to revise that position if I can see convincing evidence for it.
Next lets address the part of the claim that relates to intelligence. That there is something you can do to engineer the outcome of smarter children. Are the Polgar sisters smarter or are they better at chess? We have it on excellent authority that they are very, very good at chess. Lazslo's book is called "How To Raise a Genius!". Are they geniuses? Does it require a genius to play chess? Does it require a genius to become a chess grandmaster? Are the average life accomplishments of Chess Grandmasters in line with what we would typically call genius? Are you a genius if you can operate in the very narrow domain of moving plastic toys around a wooden board?
Not to diminish the accomplishment or belittle chess players. My point with that is I think there is a conflation between genius and expertise in a very narrow domain. We see natural born geniuses go on to pick a certain field and produce results in it, but we don't normally see parents training kids to be geniuses in a particular field.
Their story is one of my favorites and I think about it often. But you have to ask a few questions... does it generalize to different domains? Maybe it does. Maybe it doesn't. I suspect it does. Can it reliably be repeated and systematized? Perhaps, but I suspect not. Is that at all practical and useful to the average parent? Probably not?
I actually think his experiment and it's result make a lot of sense. He trained them from the age of 3. They fought hard with the government to be allowed to home-school them so they could focus on a curriculum of their choosing, so they had control over what the girls were required to do academically. They made them practice hours and hours a day. The girls, as I recall, loved it. They loved chess. That helps. Lazslo also likely devised good learning strategies. It was a completely unorthodox experiment.
Overall I'm not surprised he got those girls to where he did. I also think that as a pedagogical method it is stunningly impractical for society as a whole to try and model, so I'm not sure it's useful even though I actually think he's right on the money. The traditional education system is not really designed for education, it's designed for credentialing. You have to focus on the lowest common denominator in order to meet the required throughput with a given consistency of results. If you wildly adjust the parameters of the system to say OK what about if we only need to educate three people and we can choose the design goal of the education system? You can certainly wind up with the result that he did. Does that make them smarter? Or does that just make them hyper-capable in a narrow domain?
I think back to my own educational experiments over the years with regards to language learning and compare my results with the results of learning languages inside the university system. It's worlds apart in terms of results. I can personally do it 3x faster on average. I'm undertaking a new experiment in 2019 to deeply learn psychology. We'll see where that winds up. Suffice to say you can most definitely get out-sized results with regards to education if you're in control of the system, you're very capable at devising novel learning strategies and you aren't required to mass-produce results.