Consider the idiom it takes two to tango. Now, one of the two could be a quadriplegic, or a dog, or maybe they both come from a severe religious cult which has forbidden dancing and just can't bring themselves to do it.
That's fine; the meaning is that two are necessary, not that two are sufficient.
But to the topic at hand, the logical implication is that "to do it" => "takes surprisingly little time" (that is, doing it implies you took surprisingly little time)
Not "taking surprisingly little time" => "to do it" (that is, taking surprisingly little time implies you do it; after all, you might have taken surprisingly little time to do -something else-).
Same as the above; the implication is that "success" implies "(at least) a surprisingly small number of adults were involved", not that "(at least) a surprisingly small number of adults were involved" implies "success".
I find it helpful to use formal logic to explain stuff like this. If I make a statement that p => q (that is, p implies q), and that statement is true, it does NOT necessarily follow that q => p (that is, q implies p) is true (it may or may not be; it is independent from a formal logic perspective). It is necessary to evaluate q => p separately, in which case we may find it to be true, OR we may find q =/> p to be true (that is, q does not imply p; or put another way, the fact q is true tells us nothing about whether p is).
As an example, "My shoes are wet because it is raining", p = it is raining, and q = my shoes are wet. p => q means if it is raining (p), my shoes are wet (q). But it does not also follow from that that q => p; that is, just because my shoes are wet (q) it does not mean it must be raining (p; I might have, after all, been out watering the lawn). Thus, q =/> p.
For the tango example, p = a tango is possible, q = there are two people. p => q, but q =/> q (that is, two people does not imply a tango is possible).
In this case, p = foster child is successful, and q = a number of adults (sometimes a surprisingly small number) were involved with them. p => q, but q =/> p.
Saying "it takes a surprisingly small number of emotionally stable adults to help kids " is emphasizing that it does not require a large number of stable adults and therefore a small number is sufficient. How many does it take? Surprisingly few. A few suffice. How else can you read this? You're surprised that a necessary number is small? But it's always necessary to have at least 0 stable adults and that's as small as possible - you can't get surprised that it's small! Instead you must be surprised that the sufficient number is small.
I'm not making any comment on the number of adults; I'm just responding to the original argument that "well, I know foster kids who haven't been successful, who had supportive adults", and the ensuing argument, to further explain why that isn't a negation.
It takes two to tango is about the tango going to happen, not the minimal required circumstances for it to happen if all other stars align.
https://en.wikipedia.org/wiki/It_takes_two_to_tango
The meaning of this expression has been extended to include any situation in which the two partners are by definition understood to be essential (...).
more generally, when someone says something you find incredulous, consider processing that internally, and then decide if saying "huh, now i know" is not a better reaction to "no way, maybe you're wrong."