Is this a general question using this as an example, or is the question specifically designed to resolve this scenario? I hope it's the former, because the latter is boring.
Humans are problem solvers, or computers in the sense that we compute solutions based on the information available. Part of computing solutions is defining the boundaries of the problem, and this is where one will find the origin of this "hypocrisy" - the parents, in this scenario, have defined the risk value of horse riding per hour to be outside the bounds of the problem of motorcycle riding safety. If I were to hazard a (largely meaningless) guess, I would say that they simply don't consider horses to be comparable to motorcycles.
If humans didn't do this, it would be virtually impossible to assess risk for anything at all! Imagine if, before one had to make any risk decision, one would have to compute an optimal (rather than a satisficing) solution in the way you are looking for the parents here to: it would be virtually impossible due to combinatorial explosion. The only viable computational strategy at that point would be to avoid all perceived risk above X (fairly low) threshold, which would probably mean any behavior beyond bare necessity for survival would have to be avoided - not to actually prevent risk, but to prevent having to compute the risk.
In sociological terms, hypocrisy is "bad" because it makes it computationally harder for other actors to correctly predict the intent and future behavior of hypocritical actors, which requires more information to be processed. However, this is a great trade-off for the individual brain, because the extra information processing needed to compute solutions to these irregularities of other actors is much smaller than computing optimal solutions for every little risk problem, partly because employing limited boundaries (even though they may lead to hypocrisy) affords the human brain the capability to easily compute the risk that this will happen, which would be impossible or unfeasible without such a capability.
The motorcycle child MC is also "hypocritical" (remembering that hypocrisy is really just an observer effect) in the same way as the parents here. MC has defined the safety problem in terms of crashes per hours used. However, it's entirely possible that the average risk is actually higher for motorcycles, such as if motorcycles are used for more hours/day on average than horses, which is quite likely. On the other hand, MC might well be a motorcycle riding genius that, if allowed to ride, would never crash due to extremely well developed balance.
At what point can one objectively stop and say "this is the boundary of the problem"? It can't be done. Combinatorial explosion would leave MC calculating the problem until MC's death, and then MC can't ride a motorcycle anyway. This is why in persuasion, it's impossible to get a person to concede hypocrisy without first changing his problem boundaries, at which point conceding hypocrisy becomes moot to the persuader (since the target's problem definition is the problem).
Interesting question.