Is masturbation an effective intervention?

Note: This post is satire.  (However, I have not made any deliberate mistakes in the reasoning or computations.)

This post is an attempt to evaluate the effectiveness of masturbation, as performed by individual effective altruists.  “Effectiveness” here is meant in the usual sense of utility (as measured in, e.g., QALYs) gained per dollar invested, as used for instance by GiveWell to evaluate cost-effectiveness of charities.

Of course, as masturbation typically has no effects on others, it cannot be considered an act of “altruism,” effective or not.  However, although utilitarianism assigns no greater moral weight to oneself than to others, it also assigns no lower weight to oneself than to others.  Hence, insofar as effective altruists are interested in increasing global utility by any means, self-pleasuring acts have no special status and can be directly weighed against acts that benefit others.

To evaluate the effectiveness of masturbation, we must determine both its cost and the utility it adds.  First, consider the cost.  As there is no “price to entry” for masturbation, we can only evaluate the cost as an opportunity cost – say, the wages lost if an episode of masturbation is substituted for an episode of paid labor with the same duration.  Typically individuals do not directly face this tradeoff, but in some cases they may, for instance if one has the option of taking on additional paid hours at the cost of time that would otherwise be allocated to masturbation (or vice versa).

This post involves enough uncertainty and enough distinct data sources that only order-of-magnitude estimates will be attempted.  We will assume a salary of $80,000/year, and while this may be far too low or too high for any given individual, it will not be off by many orders of magnitude.

Thus the cost of a masturbation episode, for the purposes of this analysis, is simply its duration times ($80,000/year).

What utility is gained in an episode of masturbation?  As a first approach to this question, consider the overall utility difference a non-asexual person would incur if deprived of their sex drive.

An approximation to this question was investigated in Wilke et. al. 2010, in which men with prostate cancer made tradeoffs involving a treatment which could extend their life at the cost of “profound lack of sexual desire and erectile function.”  The men (mean age 72) were given a time trade-off question, as is standard in determination of QALY weights.  The mean time trade-off utility was 0.78, meaning that a year with sex drive and function was valued at 0.78 years without.

In other words, time spent with sex drive and function is 1/0.78 ≈ 1.28 times as valuable as time without.  (Obviously, these results are severely limited by sex, gender and age; we will treat them as universal here as a first approximation.)

The utility gained from sex drive and function is not uniformly distributed over time; it is primarily concentrated in the sex act itself.  There may be other utility gains from mood and health effects of sexual desire and activity and from the sexual drive as a contributor to social well-adjustedness, as well as utility losses due to the difficulties involved in seeking sexual activity.  However, it seems intuitive that the overall effect of sexuality on human preferences is dominated by the desirability of the sex act itself rather than by these peripheral effects.  So will we assume that if a given unit of time “with sex” confers more utility than the same unit “without sex,” this is due solely to the subsets of this unit in which sexual activity is occurring.

For instance, if a year “with sex” is 1.28 times as good as one “without sex,” sexual activity itself must be much more than 1.28 times as enjoyable as the average activity, since one typically spends only some fraction of a given year masturbating or having sex.

How large is this fraction?  Reece et. al. 2010 reviews data for men found in the 2009 National Survey of Sexual Health and Behavior.  For consistency, we must consider data for men of ages comparable to those studied in Wilke et. al. 2010, which in this case we will take to mean the “70+” age category.  Some of the data is shown below (we have chosen to exclude “anal intercourse,” which is rare enough in the 70+ age category to be negligible):

Averaging over these data, the average man of age 70+ masturbates about 24 times per year and has sexual intercourse about 20.3 times per year.  (Note that these rates may be different for men with prostate cancer.  We will ignore this difference here.)

Survey data indicates that penis-in-vagina intercourse (i.e. “vaginal intercourse” in Reece et. al. 2010) lasts around 6 minutes on average.  We will assume that masturbation episodes are also 6 minutes in duration.  This implies that the average man of age 70+ spends 144 minutes per year masturbating and 122 minutes per year having intercourse.

It is commonly observed that masturbation and sexual intercourse are not equally desirable.  Thus we introduce the parameter μ, defined so that μ minutes of masturbation are interchangeable with 1 minute of intercourse.  The total time spent in sexual activity, in units of “equivalent minutes of masturbation” are thus 144 + μ*122.

A year “with sex” is thus made 1.28 times as enjoyable as a year “without sex” solely by the contribution of 144 + μ*122 minutes, which are some factor β times more enjoyable than their equivalents in the year “without sex.”  In the unrealistic limit μ = 1, this gives β of about 558.  With μ = 5, β lowers to 197.5, while with μ = 10, β lowers further to 109.7.

As stated earlier, we take the cost of a masturbation episode to be ($80,000/year) times the episode’s duration.  To make the computation simple, consider a hypothetical year spent masturbating.  Thus $80,000 is lost, but the year confers utility β*(one year) rather than 1*(one year).  For β = 558, for instance, this can be interpreted as 557 years of life gained.  Thus we would spent $143.6 per year of life gained.

The above estimate corresponded to the unrealistic μ = 1.  With μ = 5, we instead spend $407 per year of life gained, and with μ = 10, we spend $737 per year of life gained.  Increasing μ further will of course produce even lower estimates of efficiency, but very large values of μ are likely to conflict with the results of introspection.  (We encourage readers to estimate their own value of μ, then perform the computation themselves.)

Converting to units of “lives saved,” as described here, gives us $4308 per life saved with μ = 1, $12210 per life saved with μ = 5, and $22110 per life saved with μ = 10.  It will be useful here to consult GiveWell’s remarks on cost-effectiveness:

We consider anything under $5,000 per life saved (or equivalent, according to one’s subjective values about how to compare other sorts of impacts to lives saved) to be excellent cost-effectiveness. We consider anything over $50,000 per life saved (or equivalent) to be excessive for the cause of international aid, as it implies more than an order of magnitude higher costs than the strongest programs.

Thus masturbation is unlikely to be “excellent” by GiveWell’s standards (for international aid interventions), but probably not “excessive.”  (Assuming μ takes integer values, it would be “excessive” only if μ > 24.)

GiveWell’s standards are possibly the most stringent in existence, meant for identifying the very best charities.  Medical interventions costing up to $30,000 per QALY gained are often considered cost-effective; by this standard, $737 per QALY (μ = 10) is quite efficient.

Recall that this post only intends to estimate orders of magnitude.  A closer analysis may reveal masturbation to be somewhat more or less effective than indicated here – for instance, it may be quite ineffective by GiveWell’s standards.  But it is unlikely to be very ineffective.

The above is, of course, merely an analysis of the average episode of masturbation, and care must be taken when applying it to the marginal episode of masturbation.  EAs who masturbate at a rate typical of their demographic category may encounter strongly diminishing marginal returns if they introduce additional masturbation episodes.  However, given the remarkable cost-effectiveness estimates given here, EAs are strongly encouraged to reflect on whether or not they have reached this limit.  If an EA considering a masturbation episode estimates that it will have an impact on their utility close to that of the average masturbation episode they engage in, they are strongly encouraged to proceed.  This choice is straightforward if there is no tradeoff with other life concerns, but the above analysis indicates that the choice may be utility-maximizing even if traded off against an equivalent time spent making money, when considered in terms of that money’s potential to produce effective outcomes when donated.  For instance, an EA who obtains an income of $80,000/year (the figure used above) for the purposes of earning to give should consider that some of the time spent earning this income could be spent equivalently-or-better on the task of, as it were, “masturbating to give.”