Forecaster spotlights: Meet Dawna and Ryan, our community mentors

Jul 08, 2022 05:10PM UTC

You may recognize Dawna Coutant (@DKC) and Ryan Beck (@RyanBeck) from the comment threads. They've been acting as "mentors" for the forecasting community for the past few months. Both are Pro Forecasters, who have been part of the program since before it was INFER. They were selected as our first mentors this season because they’re two of our most active and thoughtful forecasters with a track record of high accuracy. They regularly post interesting insights that can help the rest of the community with new information. To get to know Dawna and Ryan better, we asked them a few questions about their interest in forecasting, their approach to estimating the unknown, and more! Be sure to check out Ryan's tips on how to boost your accuracy score at the end.

Dawna Coutant

What is your day job when you’re not forecasting?

I was a psychology professor, most of the time at the University of Hawaii at Hilo. I made it to Department Chair before I decided I needed a change. My area is social and cross-cultural psychology. I also taught Research Methods.

How has your background been relevant to your forecasting?

The psychology, understanding of biases, cognitive processes, and cultural differences as well as how to evaluate research design, data, and arguments has helped refine my forecasting.

What do you like most about forecasting that you can’t get in other intellectual domains, such as reading or listening to expert thought-pieces or analysis?

The objective measurability of determining success. I used to unconsciously buy into the idea that the most confident voice is probably the right one. I have found since then that this is absolutely not the case. I want to know if my opinions are accurate and how I compare to others. I don’t want a bloated sense of self-worth, but I also don’t want to defer to the loudest voices. I like the model of putting our forecasts out there, and then seeing how the chips fall out after the question closes.

What are your tips for new forecasters that are trying to become more accurate?

Read other people’s rationales and ask questions if their rationales don’t make sense. Nicely, of course. Pay as much attention (and maybe more) to those who disagree with you than those that agree with you. Ask your other forecasters for a second pair of eyes if you are seeing/interpreting something and you aren’t sure you have it right.

What was your most humbling experience in forecasting? Your proudest moment?

Most humbling was forecasting the Republican Party Presidential Primary in the 2016 election (hangs head in shame). My proudest moment was either the IARPA Hybrid Forecasting FOCUS challenge, when our small group of forecasters (@Cmeinel was on the team too!) won the hybrid competition, or GJO2.0 Covid-19 Forecasting Tournament. I was in the Top 10 initial forecasts and second on final forecasts.

Besides forecasting, what are some of your hobbies?

Skiing, travel, walking my dog. And I write mysteries.

Anything else you want to share with the community?

I’m always impressed by the people I find on these forecasting platforms. People I would never have met in my normal daily life. I would encourage you to reach out and interact with some of your forecasting colleagues. They might have a great taste in music, or live somewhere you’ve always wanted to visit.

Ryan Beck

What is your day job when you’re not forecasting?

I'm a bridge engineer with a BS and MS in Civil Engineering.

Have you found your career or education to be relevant to your forecasting?

Not directly (nobody asks questions about bridges or engineering and there aren't really any interesting bridge questions to be asked anyway), but the skills I learned in my career and in college have definitely been helpful. I work with Excel constantly at work, and being able to open up Excel and manipulate large amounts of data is super helpful for forecasting.

What was the most difficult part about learning to forecast more accurately?

I think it was recognizing when I'm being too optimistic. I've noticed that my forecasting misses often involve underestimating bad developments (inflation and coronavirus, for example) and overestimating the chances of progress being made (for example, a bill making it through Congress or policies I view as detrimental being removed). I think I've become better at addressing that tendency toward optimism in my forecasting, but it definitely takes a lot of reflection.

What do you like most about forecasting that you can't get in other intellectual domains, such as reading or listening to expert thought-pieces or analysis?

I love that it focuses entirely on testable and measurable predictions and the effect that has on the quality of discussion. The incentive to be right can often moderate our opinions, it makes us pay attention to the risk that we're wrong and the reasons we might be wrong. If we ignore those risks we usually end up with overconfident predictions. Pundits often make overconfident predictions and suffer no consequences for them, but with forecasting we're assigned a score and over time our score will punish us for overconfident predictions. The feedback from scoring makes us more accountable for our errors, incentivizes us to consider other points of view and the risk that we're wrong, and builds a track record that others can use to assess how much credence they should assign to our forecasts. Couple that with the precision involved in forecasting (saying something could happen is vague to the point of being meaningless compared with saying something has a 25% chance of happening) and it seems pretty clear the benefits of forecasting are immense. The value of forecasting is something I could rant about for a long time, but I think that covers the key points!

What are your tips for new forecasters that are trying to become more accurate?

One thing I think that's really helpful is to describe your rationale in detail. Your rationale isn't just to persuade others or provide information to INFER, it's a diary of your thought process that you can look back on and see where you went wrong. If you get a bad score, look back at your rationales and see what you missed. We often have a tendency to see ourselves in a favorable light, so if you enter blank or undescriptive rationales it can be easy to see a bad score and say "well I got a bad score but it was just bad luck, my reasoning is still sound." But if instead you provide rationales that explain your reasoning, you can often see where you were right or wrong.

What was your most humbling experience in forecasting? Your proudest moment?

My most humbling has been trying to forecast coronavirus. I didn't expect the Delta and Omicron waves to be as large as they were, I thought coronavirus was more likely to recede with maybe a few smaller waves and variants. This was definitely a case of being too optimistic. My most proud are the questions where I did some detective work and it paid off. For example, on a question about some military testing, I reached out to a Defense journalist on Twitter, and she helped clarify some of the reporting I had been seeing. Doing a deep dive into a question's data or background and finding important information is a rewarding feeling!

Besides forecasting, what are some of your hobbies?

My hobbies vary a lot, but one constant has always been reading. For fiction I love sci-fi, fantasy, and thrillers. Aside from those and spending time with my wife and daughters, my other hobbies include video games, writing, programming/web design stuff, and board games and D&D.

Anything else you want to share with the community?

I'm always happy to talk about forecasting or anything else, so if you have any questions, want to talk about a forecast, or think one of my forecasts/rationales is way off base, please reach out!

Ryan’s tips for boosting accuracy:

  1. Update often: Stale forecasts have often resulted in some of my poorer scores.
  2. Engage with the community: Comment on other rationales and ask people to elaborate on things you're curious about. People are often friendly and you can learn a lot doing this.
  3. Learn software to help with data: There are many times when knowing how to look at and plot data will help you make a better forecast. Excel or Google sheets can be extremely useful for this, and if you're feeling up to it consider learning R or other useful programming languages. Not only are these helpful for forecasting, but being skilled with these can help in many careers as well.
  4. Be a skilled searcher: Learn to use the tools on Google or other search engines that let you filter by date, this can be helpful for seeing only recent stuff, or excluding recent developments to see what's happened in the past. You can also set Google Alerts to receive emails about topics relating to questions. There are also tons of scientists, economists, and journalists who care about accuracy on Twitter, and you can learn a lot there if you fill your Twitter feed with people who care about science and being accurate.
  5. Read Thinking Fast and Slow by Kahneman: This is a super helpful book for understanding the importance of base rates and recognizing our biases and other pitfalls common in forecasting.
  6. Recognize limitations with metrics: Many measures have issues. The way we measure and quantify the world is often going to be imprecise or suboptimal in some ways. Understanding the methodology behind the metrics being used can often help make hard questions easier to forecast on. And that doesn't mean the metrics are bad or not worth forecasting on, often these same metrics are reported on in the media, so being more informed about caveats and limitations is valuable and can make you a more informed consumer of the news and a more informed forecaster.

Be sure to follow our mentors @DKC and @RyanBeck to keep up with what they’re posting and sharing! 

Tip: Mention someone by typing @username