Those of us of a certain age might recall when Barbie could utter the phrase “Math class is tough.” A good example of this is how to figure out the data transparency in the time of the Covid.

One of my go-to sites is the Covid Tracking Project, which is a group of computer scientists that daily scrape and interpret the thousands of county health stats for testing and infection data. You might have noticed that for each state’s data summary they issue a letter grade for transparency. How they arrive at that grade is instructive, and we should all take a moment to understand the calculations. Even if our business isn’t involved in public health, it can help inform and improve our own transparency efforts.

Just look at some of the recent transparency disasters from last summer, when Facebook and Equifax couldn’t be trusted with showing the truth behind their numbers. We want to be more transparent, because that means we have the ability to create trust with our customers and partners. So let’s look at how the Covid Tracking Project assigns these grades to each state and US territory.

Their transparency grade uses16 different metrics. These include factors such as: is the state’s official health website the best data source and consistently updated? Does the state report patient outcomes, such as how many patients are on ventilators? Does the state break down the demographics into ethnicities, race and pre-existing ill patients? How about total hospital capacity for the state? For each metric, the data quality can vary and the details matter. For example: some states just report positive tests and deaths. For some states, you have no way of knowing how many negative tests were obtained, or how many of those who tested positive then went on to consume an ICU or ER bed or other hospital resources.

The transparency grades are calculated each day: I have noticed that the grade for my state, Missouri, has varied from A to C. Today Nevada, Nebraska and Puerto Rico all have failing grades.

But wait, there is more. The project team also has a Slack channel and a GitHub public project where you can dive deeper into what is going on here. The former is used to address reporter’s questions and the latter is used to call out support or bug issues. The team also has taken pains to explain exactly what they are counting — for example, they look at where people are being tested, which is not necessarily where people first became ill. Every state reports these numbers somewhat differently: some use online dashboards or hyperlinked data tables, while others announce their stats at daily press conferences or via social media posts. The team has taken pains to double-check everything and annotate where things are ambiguous or unclear.

I should mention that the project relies on dozens of volunteers too: so managing all this collaboration is key. Clearly, there is a lot we all can learn from their excellent transparency efforts.

Now that we are sticking close to home, we are using web conferencing tools. No matter how tech-savvy you might be, running a great conference isn’t easy and will require a collection of people with various skills: part TV producer, part sound engineer, part professional speaker, and all sprinkled with a good deal of patience and troubleshooting. For the past several months, I have been on the production team of a rather large conference for the American Red Cross. We have a team of more than a dozen people that puts on this event every month, and lately has had several hundred attendees and multiple presenters. Every month we find and fix new and interesting problems, some technical, some social, some particular to Webex.

Before I give you some lessons learned from this and other web conferencing experiences, I want to relate an anecdote from this month’s call. I was talking to one Red Cross volunteer who was having trouble getting connected. When he told me that he was at a Red Cross blood drive and actually giving blood, I suggested that maybe he should just wait and watch the recording of the event. A few minutes later, he emailed me and told me that he had figured out how to tune in for the meeting. That is devotion!

Here are some suggestions so hopefully you can make your meetings more valuable and professional. You might also want to review another blog post that talks about more general collaborative techniques. 

• Decide whether you want to display everyone on camera, run a live demo from someone’s computer or focus on the slide presentation. You can’t really do all three well, and switching from one to another can introduce issues. Pro tip: if you are sharing your screen, you’ll want to share it from a second monitor that is using a lower resolution. 
• Simplify, simplify. Eliminate options to help reduce user confusion, and simplify the technical details whenever you can. Keeping things simple means less to go wrong. For example, we use two chat channels: the one built-in to Webex is for the attendees to ask questions (we don’t use the Q&A feature to keep things simple) and we have separate Microsoft Teams chat sessions so the production team can communicate with each other as issues arise during the session.
• Use a consolidated slide deck. Give a deadline when additions/changes will be accepted, thereafter the deck is locked. This means someone will be the lead producer, who will advance the slides for everyone. On Webex, there are three different ways to share your slides: “Share the actual PPTX file” (recommended), “Share Application,” or share your entire desktop. Pro tip: with this latter method, you will want to turn off notifications on your computer — Focus Assist settings should be turned off in Windows or Notification Center set to Do Not Disturb with Mac. 
• Assign roles with care. The more you can segregate these roles and spread among different people, the better the overall experience for your attendees.It also allows the production staff to focus better and provide a better attendee experience. We typically have at least six people that run each webinar:
  • the host, who acts as a master of ceremonies and keeps everyone on schedule.
  • The lead producer, who sends out meeting reminders and calendar invites, advances the slides, does audio checks, and is charge of everyone else during the event. The producer also posts a recording of the meeting and chat sessions to the various online repositories.
  • The secondary producer, who troubleshoots problems for attendees and presenters and responds to emails during the event about production and connection problems.
  • Two chat monitors: one as backup, one who will read the questions aloud and direct the appropriate person to answer them during the meeting at specific points.
  • A graphic artist, who whips the slide deck into shape visually.
  • In addition to these roles, there are several others that work in the background. We all report to a team leader, who is ultimately responsible for organizing each month’s speakers and does final review of all of our materials prior to the meeting. We also have note-takers who organize the after-action report and follow up with any promises made during the meeting.
• Rehearse. All presenters should have a sound check prior to the start of the meeting. We usually do this the hour before the meeting: given that we have so many presenters, we want to make sure they can be heard clearly. This is usually where the problems happen, so resolving these early is key. We usually recommend to use a headset with its own boom mic and not to use a speakerphone. Also, rebooting your PC just prior to the meeting is a good way to clear out problems. Presenters should also rehearse moving from slides to live screens and back: the transitions can be tricky under certain endpoints and web service providers.
• Put together several documents to help your production staff: these include a sample timings sheet with speaker, topics and start and ending times, a contact sheet of everyone’s phones and emails, and other production details.

As you can see, there is a lot of work to produce a quality web conference. Feel free to share your own tips here as well.

With all the talk about “flattening the curve” and model disease predictions, I thought I would take the opportunity to explain exactly what these models look like. No, they don’t bear any resemblance to those people that walk down runways or the plastic things we put together with that smelly glue when we were kids.

My first brush with the art and science of math modeling was in graduate school at Stanford. I actually took an entire class in how to build them, and it was one of my favorite classes. Each week we would have a new assignment. What I remember about this class is that the assignment could take anywhere from a few hours to several days to complete. Sometimes — most times — the assumptions that I made were so wrong that I would have to start over. I remember a few of the assignments: one was to build a model to help women decide when they should get screened for breast cancer. Another was to help the Palo Alto school system decide which of their grade schools to close for under enrollment. These two were very tough, and I can’t remember if I came anywhere near the correct approach, just that they kept me up for many late nights.

What made this class interesting was that it was taught by two business professionals who lived in the “real world” and deliberately chose examples from some of their consulting projects. That particular class is no longer being taught. Indeed, searching through Stanford’s website, I was chagrined to find out that my degree, Operations Research, was eliminated more than 25 years ago. Time marches on.

Actually, I had an earlier experience with math models. I had a student job rebuilding these antique brass and string models that were created in the 1830s. They were used to teach students how to draw conic sections, back before we had Mathematica or even full-color textbook illustrations. But these are more literal “models” — for this post, I am talking more ephemeral constructs that are mostly data and equations.

Math models are being used all the time and don’t usually get much attention — until disaster strikes. One type of them that you consume daily is weather forecasting models. When a hurricane threatens part of the world, you see a variety of forecasted paths that the storm is likely to take: each one of those paths comes from someone’s model about past behavior. Another example is building a model to calculate how much flooring you’ll need to cover your room. That one is pretty simple, using additional and multiplication, but still you have to do the math to figure out whether you’ll need 10 or 20 boxes of materials.

Some of the most annoying math models were those “word problems” that we all had to solve in grade school. Maybe that is why many of us have steered clear of them in adult life. But after taking that modeling class in grad school, I got into math modeling as a career, and went to DC to become a consultant myself. I built models to support public policy decisions, such as whether to build a dam threatening an endangered species (my model said to go ahead, based on the economic outcomes) and whether to enact building energy conservation standards (my model said yes to that too).

Here are two places to look at noteworthy virus modeling efforts:

• Unacast’s Social Distancing Scoreboard. This shows the change in average distance traveled across each county, using anonymous cell data. The current data is compared to before the outbreak, to see how we have stayed closer to home (or not).
• Washington State doctors and the Institute for Health Metrics and Evaluations have created a model that predicts the peak of infections across the world. The model is updated daily and has information for each country and then by state that is used by hospitals to plan their treatment efforts.

Stay healthy and safe wherever you are.