When You're Asked to Solve an Impossible Problem

What can you do when someone asks you to solve what is clearly the wrong problem? It’s a common dilemma, and how you choose to deal with it can have serious professional and personal consequences.

Sometimes you can see there is a better problem that would deliver the desired outcome with fewer trade-offs or resources, but other times you are being asked to do the impossible.

In those cases, I follow a three-stage process. In the first stage I check my understanding of what the person with the problem is trying to achieve. Second, I explore whether there are better ways to solve the problem.

If both these options fail, then the third stage becomes a test of my personal values, essentially asking “how much compromise am I prepared to accept, personally and professionally?”

Borrowing from How to Speak, How to Listen by Mortimer Adler, your first job is to understand what the person is saying. In this part of the discovery process there are a few major things I’m looking for:

• Am I talking directly to the person who has identified the problem? If I’m not, I now need to decide how much information is being lost in transmission, and whether I need to work my way upstream to get clarity.

• What would they like to be different about the world after the problem is solved, or what would be different in the future state compared to the current state?

• Why do they think this is a solvable problem? I’ve recently been asking for their evidence with good success.

Next, I start to explore the likely tradeoffs. Maybe the problem is clear and straightforward, and I think we can solve it. Great! But a lot of the time there are some requirements that will be expensive and useless.

For example, in analytics, a favorite request is for “real time data,” but often for a report that will be viewed once a week. This adds possibly substantial costs for a negligible improvement in accuracy. Some questions I ask:

• Why do you want expensive requirement? This is always specific to the problem we’re solving, but I use my expertise to estimate what that requirement is worth to them.

• Would this alternative cheaper requirement fit your needs? This involves guessing what a reasonable substitute might be. Even when the answer is no, it can help give more clarity on the reasons they asked for the expensive requirement.

• What if we could do this other thing instead? Sometimes the whole framing of the problem seems off, and so I test if there are other things they might accept as a replacement problem.

Lastly, I assess the unknown by applying sensitivity analysis in the problem-space. This might happen days or weeks after the first two phases when we’ve started to solve the problem and are seeing how things are shaping up. The questions are basically the same as above, just after I know more.

Following the above process, I can typically shift the problem to something better. It might not be perfect in my opinion, but at some point I accept that I’ve done my part and there just might be some things I don’t know that make the current direction the right one.

Sometimes though, the person who gave me the problem is still determined to solve something I disagree with. At that point, I shift my focus to choices based on my personal values and beliefs. My options might include the following:

• Deliver exactly what I’ve been asked to with a minimum of effort. This allows me to hopefully not face negative repercussions without wasting more of my limited hours than necessary.

•  Deliver something different than what I was asked, because I think when they see the result, they will finally come around. This is often the best choice but is highly dependent on the situation.

• Refuse to solve the wrong problem. This may result in getting fired in some cases, but at times it’s the right call.

• Try to prove through my work that the problem is wrong. This is both risky and difficult. But it may be your only option (except to refuse) when you’ve been asked to do something that is actually impossible.

All of these options pose deeply human dilemmas, and all entail some personal risk. Only a human can make this transition from “how do I do this” to “what should I do?” “AI” or large language models don’t know when they should stop and computers do what they’re told. But people can enrich their decision-making with values, ethics and moral codes. Maybe that’s what equips us to deal with impossible problems.

One of my core beliefs is that we have the power of choice in everything we do. We should use it.

My manifesto covers some of these ideas and is now published on www.meta-problem.com/about. I welcome discussion.