Jakub Wiśniewski argues, contra my  claim, that there is a value-free way to define reference classes for the BST. Reference classes, Wiśniewski proposes, are natural classes of organisms whose shared features do not reduce their prospects of survival and reproduction below the statistical average for the entire species. Though this good suggestion can accommodate several counterexamples, I do not think it can succeed fully.
First, compare survival and longevity for women and men. In nearly all human societies, women can expect to live longer than men and male mortality exceeds female mortality at all ages including – in countries with good health care provisions – the childbearing years. This means that being male – a natural class if there is any – reduces the prospects of survival and reproduction below the statistical average.
One might respond by supposing a considerable reduction needs to be achieved: more than the year-or-two below average reduction that being male results in. But that only raises a dilemma resting on a well-known problem: some diseases, such as the common cold, don’t reduce one’s prospect of survival much [if at all]. Therefore if a considerable reduction in survival prospects is necessary, this would allow for a reference class of people that have the common cold, as well as other diseases that do not significantly impact longevity.
Second, Wiśniewski supposes that ageing is not a counterexample, because although different age-groups have different survival prospects, some of which come out below average, this is not caused by age. But that is a not a valid response.
First, it is unclear that the lack of a causal relationship holds; whilst the passage of time may not cause a weakened organism, the increased strain associated with the passage of time, such as reduced telomeres and free radical exposure, does – and it is that which age picks out.
Second, and more importantly, by demanding that the feature uniting a reference class must cause the reduced survival prospect, Wiśniewski’s account opens the door to admitting groups that can be identified by features other than the one that marks their decline. An example is the natural class of people whose urine contains glucose. Since it is not glucose in the urine that causes their symptoms and reduced survival, but the underlying diabetes, people with glucose in their urine is an admissible reference class if people over 70 is in Wiśniewski’s account – even though “people with diabetes” or “people with reduced telomeres” are not. Thus, inadmissible reference classes can be resurrected: Down’s syndrome can be “people with an epicanthic fold”, hepatitis and other liver failures “people seeing yellow”, and so on.
If, on the other hand, we drop the causal link between the features that identify a group and its prospects for survival and reproduction, and just took the group’s survival prospects at face value, ages would no longer result in admissible reference classes. This is a problem because it is not just the very old that have increased mortality and reduced reproduction; it is also the very young. And where someone might be willing to bite the bullet and consider ageing a disease, this is not an option for newborns.
Boorse called reference classes into action precisely because these groups have differing survival and reproduction prospects, but each are healthy. Appealing to overall group survival and reproduction in comparison to the species averages will therefore not generate admissible references classes, whereas appealing to causal features within the group is too open for redescription. Some other way of picking out reference classes is necessary if Boorse’s account of disease is to be salvaged.