Managing Difficult Decisions in Medicine with the 40-70 Rule | Hacker News Reader

Managing Difficult Decisions in Medicine with the 40-70 Rule

paulsufka.com

32 pointspsufka11 comments

It's interesting that they bring up Colin Powell in this, particularly because the big 'flaw' with this type of advice is lies in something that Donald Rumsfeld is infamous for saying - his so called 'known knowns' (or conversely, unknown unknown) phrase.

You're ability to even judge that you have 40 to 70 percent of the information required isn't even a given. In particular, especially with something like medicine, where there is time pressure, bizarre individual circumstances, confounding issues, it can make it difficult to really compress your current state of knowledge into a single value representing percentage of 'known' information.

That said, hilariously if you look at the reproduced slide in the bottom, the blog post and the slide actually say different things.

Colin Powell says to act when you have between a 40 to 70 percent chance of being right given your current state of knowledge/information. The slide says acts when you have 40 to 70 percent of the necessary information.

The article is copy-pasting what seems to be a widely spread misquote of Colin Powell. What I think is the actual quote seems less controversial than the "What defines 40% of the information?" issue that is being debated here.

> Part I: "Use the formula P=40 to 70, in which P stands for the probability of success and the numbers indicate the percentage of information acquired." Part II: "Once the information is in the 40 to 70 range, go with your gut."

So, Powell was actually talking about confidence in success, not percentage of information. Confidence is explicitly an estimate, whereas "percentage of information" is clearly unknowable in the situations being discussed.

The original collection of Powell quotes that included this one is a lot more interesting and valuable than this particular article, IMHO. http://govleaders.org/powell.htm

This is not always a good approach. It can only be used as a rule of thumb. Here is a counterexample:

I had to go to the hospital because of extreme pain. The doctors were pretty sure that I have some kind of auto immune disease which could have been treated with immunsuppressive drugs. They artificially delayed the treatment. If they had started the treatment immediately I would have felt better sooner but also some symptoms which only manifest itself after a certain amount of time would have never been seen. So for a long time we only had a imprecise diagnosis. It would have been enough to treat it but it would have also hidden many symptoms which were needed for a more precise diagnosis.

Especially if you have a chronic disease you want a very precise diagnosis because you will be treated for the rest of your life. If the treatment is not as effective as it could be you will have problems in the long run.

So they gave me very strong pain killers and let me rot for 2 weeks or so. Then the treatment began and it took another 3.5 months once I was released from the hospital.

I am happy that I now have a better near optimal diagnosis.

Where do the '40' and '70' come from? Who's defining what the final amount of available information is, and judging what the current percentage is? How do you even know at the time?

I mean, it would be one thing to create some kind of simulation and discover that this was actually something statistically meaningful. But there aren't exactly any citations here..

Your second question answers the rest. There is no way to accurately give numerical values in real world applications, so 40 and 70 are chossen as mental model approximations. The fallacy of drawing the line is making no choice or distinction because there is no method to determine the correct one.

It's really just pseudo-scientific hand-waving to make people feel better about decisions that they make.

As you noted, if this was something more than that there would be simulations, citations, examples, or really anything, other than feel good puffery.

Well, if you think of it as some generalized instance of the secretary problem, the 40-70% rule wouldn't necessarily be that bad --I guess it depends on how adversarial are your assumptions [1].

[1] http://math.stackexchange.com/questions/33227/3-other-extens...

Without any data behind this, it's just one more wacky theory of management that is being promoted without skepticism.

It's a bit like "brainstorming", the strategy of idea generation that has an appealing narrative, but doesn't result in greater creativity when measured objectively.

I have to admit it's not what I guessed it was. From the title I thought it would be something like "under age 40, treat aggressively, over age 70, just take away their pain."

Of course a decision can't be made with 0% information, and waiting for 100% information in unnecessary. But how would one arrive at the range of 40-70%? For a decision about risky surgery I would hope that the amount of known information is much closer to 90%.

"40-70 rule" seems like a banal remark once you realize different decisions require different amounts of information

If the surgery is truly risky, it may be impossible to know even 70% of the information. Risky surgery tends to be rare, and the people who need it tend to have more than just one problem, which compounds the number of variables that can affect the outcome.

Reply on news.ycombinator.com