The GiveWell Blog

How Do You Value Information?

There’s a common piece of dating advice: before you commit, go on a big trip together. Hopefully the trip itself will be fun, but that’s (almost) beside the point: the real goal is to figure out how smoothly you solve problems together. When it starts raining at 10pm on a Friday and you’re caught outdoors because of a misread-map incident (if we’re being honest: entirely your fault), you’re going to learn a lot about whether your partner reacts to difficulties in a way you’d like to be around for the rest of your life.

The value of what you learn from the trip (a concept called “value of information,” or VOI) can far exceed the direct costs and benefits, because the trip is just for a week and the rest of your life is, well, the rest of your life. If there’s a 10% chance that the trip will show you that you’re not truly compatible, the expected value of that information is higher than any plausible level of (un)happiness you might experience on the trip itself.

Sometimes, the value of information makes the trip worth taking when it wouldn’t have been otherwise—your partner has already scheduled a trip during their time off and it’s very inconvenient for you to join them, but the value of information makes the trip worth it. Other times, you can make choices about the trip in order to increase the value of information you’ll receive: perhaps an all-inclusive beach resort would be more enjoyable today, but a road trip where you camp at different spots each night has a higher value of information and so is ultimately more valuable.

There are many situations in life where we implicitly make choices in order to maximize the value of information, even though we don’t always frame it that way:

  • Some internships exist largely for their value of information: the costs to the company are higher than the direct benefits of the work done by the intern, but they provide valuable information about which candidates to make full-time offers to.
  • Before buying a house, you might visit it at different times of day and night to see what the neighborhood’s like and whether the realtor’s claim that you really can’t hear the road holds up.
  • Before you agree to get your kids a dog, watching them pet sit for a weekend gives you lots of information on their actual willingness to go for two walks per day, come rain or shine.

At GiveWell, we explicitly model the value of information in many of our grants. For example, value of information is a key reason we make grants for research (like this one to the London School of Hygiene and Tropical Medicine for a randomized controlled trial of eyeglasses distribution). In other cases, we build in intensive monitoring and evaluation to provide additional information regarding programs we support. In the case of MiracleFeet, for instance, we funded an implementation grant that provides direct benefits to the children receiving treatment, as well as a grant to support intensive monitoring and evaluation, which provides additional value by helping us learn more about the effectiveness of the program.

You don’t need explicit, formal models like ours for value of information to be an incredibly useful concept; any time that you can pay a cost now that’s smaller than the long-term value you would glean from better information, it might make sense to spend a little bit of investment in information now in order to improve your outcomes later.

So, for example, if there are five coffee shops near your office and you’re planning to buy a coffee every day all year, it might make sense to start by doing a proper taste test at all five to see which is your favorite, even though that costs you additional time and money. Since this information can affect future decisions, it might be worth investing in the information up front. On the other hand, if you’re at a conference in a new city for one weekend, it’s probably not worth doing a coffee shop tour there unless you get direct enjoyment from that, as you likely will not be able to use your newfound knowledge in the future.

One last important note: when seeking information, a wide range of results can be useful. This is true for GiveWell’s grants. While of course on some emotional level it’s natural to hope that results will come back positive for any particular program, in a larger sense we’re grateful even for information that’s “disappointing”: if it turns out a program is less cost-effective than we thought, we can learn from that information and allocate funding instead to programs that are more cost-effective.

Let’s go back to the dating situation from the beginning. While of course it’s natural to hope that the first big holiday together goes swimmingly, if it causes you to break up, you don’t need to see this as money wasted on a trip with someone you no longer care for. Instead, you could view it as an incredibly successful experiment that prevented you from sustaining far greater losses in the future. And if it goes really well, you’ll be able to move forward together more confidently.

New Comment

Your email address will not be published. Required fields are marked *

*