First off, kudos for sharing your codification of a useful and rational approach to evaluating advice, which I think is extensible to plenty of other things (don't buy your parents an extremely complicated DVR, don't buy your kid a Standard Transmission if you're not planning on teaching them how to use it, you probably don't need another infomercial kitchen appliance, etc.). But you missed a very important part of giving advice: don't treat your audience poorly! For example, claiming that your version of good advice is "by definition" the better advice, or that those giving other advice are being both unhelpful and dicks.
Clearly, the difference between your idea put forth here and what you're railing against, is the difference between "best effect on average for a random sample" and "best possible effect for an individual," which does not invalidate the "best possible effect" being the best advice, because - and this is the key that other commentators have hit upon already - your advice must be judged based on who it's being given to! In general, a /. audience will be more likely to take more complicated yet ultimately more effective tech advice than the general population, for example.
So I put forth that your entry here needs a simple revision as follows; WABR should use a sufficiently-large sample size of people _similar to the one(s) being given the advice_, and when performing mental WABR estimations, remember that you may potentially need some explicit weighting based on how important SOME positive effect for many is, compared to how important THE LARGEST positive effect for a few is. The computer antivirus example is a good one where cutting back on most viruses for most people is probably more important than avoiding almost all the viruses for a few people, while in a real-world virus advice comparison between (A) Avoid crowds and sick people when you can, OK?, or (B) Go get vaccinated for stuff, dummy!, the less-likely-to-be-followed advice may still be the best to give, due to herd immunity effects.
Then there are some lesser considerations of encouraging both WABR and CBR advice-sharing, things like 1) being given both a piece of difficult advice and a piece of easier advice, makes the easier advice more likely to be followed because it seems much more achievable than when presented on its own (thus increasing the value of the typical WABR advice by acknowledging CBR advice), 2) the fact that giving CBR advice will be a positive effect for some people whether or not they're already following WABR advice because they will see its value immediately, 3) CBR advice gives people room to grow if they later decide that WABR advicewas not enough.
Which makes me wonder if we're discussing strategy for the wrong person here... maybe it's the advice _seeker_ who needs a new strategy: to ask advice of many people until they have a sufficiently-diverse set of options and frequency distributions for those options, before following any advice.