Class M-3 Model B9, General Utility Non-Theorizing Environmental Control Robot

ERROR WILL ROBINSON ERROR.

John and Maureen Robinson survived their stint on Alpha Centauri, so Will and Penny never had to languish in foster care, cringing in the corner when Johnathan Harris would simper by to leer at the boy and sneer at the girl.


Few earthborn children suffer the same fate as the crew of the Jupiter II these days, but according to the Child Welfare Information Gateway (of Children's Bureau fame), as of 30 September, 2010 there were an estimated 408,425 children in foster care. Is that a lot in a nation of ~300 million people? That's not a rhetorical question. I do not know what the optimal quantity of children in foster care is. My guess is that on average, it's better to be alive and in foster care than to have never been born at all. I also suspect that on average, it's better to have been adopted than to shuffle from home to home in foster care.

For those kids who stand no chance of being reunited with primary caregivers or close relatives, they will exit by way of either emancipation or adoption by strangers. Think of the normal live birth rate as an exogenous flow parameter, with some fixed proportion of this rate landing in the foster care system (please indulge this simplifying assumption for the purposes of this discussion; I quite well understand that finer and more complete analysis will endogenize the main flow and the shunt parameters alike). The outflow parameter is governed by the adoption rate. If a policy goal is to improve the welfare of children in the foster care system, increasing the speed and the volume of adoptions seems to be the best way to do it.

Conditional on errors, of course. You don't want kids languishing in foster care, but neither do you particularly want them adopted by abusive parents. You want arbiters of the adoption decision to avoid both Type I and Type II errors.

To review, a Type I error is a false rejection of a true null hypothesis. A Type II error is the failure to reject a  truly false null hypothesis. Since the unknown in this case is the moral character of potential adopting parents,  the selection of the null hypothesis becomes quite important indeed. Here are the two zero-information options: (a) all applicants are evil until proven innocent (and by 'evil', I mean that the expected outcome for adoption is unambiguously worse than the status quo of foster care) or (b) all applicants are good until proven guilty (again, by 'good', I mean that the expected outcome for adoption is as good or better than the status quo of foster care). This leaves us with our old pal, the 2x2 matrix, as illustrated here:

	Type I	Type II
(a) Evil	1	2
(b) Good	3	4

Four cells, four outcomes. Let's have at it.

1. Under null assumption (a) a Type I error means that applicants are actually pimps or organ farmers or slavers and they've managed to successfully fleece the gatekeeper authority into thinking they're honest, salt-of-the-earth folk. Of course, in reality, these sorts of errors probably won't be as monstrous as I've indicated here, they'd just have to be worse than the BATNA of foster care (hint hint).
2. Under null assumption (a) a Type II error means that honest, loving parents get the permanent stinkeye from the decision-making authority. The kid is forced to accept the BATNA of foster care and the potential adoptive parents have found that they flushed a bunch of time, effort, and money down the toilet only to go home empty-handed, lonely, and despondent.
3. Under null assumption (b) a Type I error means that the granting authority has completed their due diligence and accidentally found honest folks to be rapscallions. This is a little different than cell 2, since the default assumption is that anyone seeking to adopt is basically a good person. The authorizing agent needs to find actual evidence that the applicants are bad before rejecting the application rather than building a case that they're good. The parallel in the justice system is in the burden of proof. Criminal law cases require "beyond a reasonable doubt" and tort cases require "preponderance of the evidence". Cell 3 is more like an innocent defendant serving time than a non-negligent party paying damages.
4. This is the really bad one. We've got bad people who have sailed under a relatively high bar. Since the default assumption is that people seeking to adopt are expected to be good, the criteria for rejection should be difficult to meet. Clever enough evil folks should be able to meet it easier than under assumption (a). More kids will end up in bad homes this way.

Fear of cell (4) is a not-unreasonable excuse to make assumption (a), that applicants enter adoption agencies under the presumption of guilt. But what if there were easier ways to mitigate cell (4) risks that didn't involve lengthy, often torturous application processes, stymied along the way with interminable paperwork, wasteful fees, and in the case of overseas adoptions, seemingly endless streams of bribes? How about we look at base rates and investigate, then eliminate the incentives for bad people to adopt children? 

Or how about we refrain from making the Nirvana fallacy? Instead of comparing life in an adoptive home to the idyll of being raised by both parents living in a middle-class single-family home in the 'burbs, perhaps the appropriate comparison is life in a stable household headed by adoptive parents compared to the relative instability of short-term foster care?

Or most importantly, how about starting with something as simple as completely open borders in adoption? This checklist is appalling. 

I get that adoption is not euvoluntary. Kids have little recourse in the process and their BATNA could be unattractive. There's also the possibility that exchange of the basket of rights implied by legal guardianship could be abused by people of ill intent. I agree that it's wise to try to ensure kids don't end up in harmful homes, but I also worry that if we let the perfect be the enemy of the good, many potentially good homes will go empty, many deserving kids will keep their suitcases permanently packed. It seems at least plausible to me that some sort of institutional change (like lowering the baseline standards for adoption to recognize the actual opportunity costs of adoption rather than some imaginary, idyllic standard) would be both directionally euvoluntary and welfare enhancing.