Who We Are
Paul grew up in New Jersey but headed West to attend school at the University of Montana intending to become a fuzzy mammal biologist. He filled his schedule with math and computer science electives instead and hasn’t looked back. He has a PhD in fishery and wildlife biology from Colorado State University and worked as a biometrician for the Colorado Division of Wildlife for six years, where he first built individual species population models. He returned to the University of Montana in 2011 as an associate professor of quantitative wildlife biology and strives to apply mathematical theory to on-the-ground wildlife management problems. He has worked on everything from mule deer population dynamics to cruise ship collisions with humpback whales. When he’s not working, Lukacs can be found hunting, fishing, hiking or skiing—though chances are he’s still thinking about statistics no matter the season.
Josh has a tendency to make his hobbies harder than they have to be: playing hockey goalie, archery hunting and working in statistics and computer coding. He grew up fishing and hunting in Michigan and then spent five years in the U.S. Navy as an Explosive Ordnance Disposal technician, trained to handle and disarm explosive devices and chemical or nuclear weapons. He does a little less parachuting these days but still brings the same attention to detail and love of problem solving to his current role. He earned a bachelor’s degree in wildlife biology from the University of Montana and then a PhD from Université Laval in Quebec City, Canada. There, frustrated by a lack of statistical software packages capable of answering his many questions about population dynamics, he started teaching himself computer coding and Bayesian statistics and set out to build his own. He joined the Lukacs lab as a postdoc in 2012 and is now a research scientist at the University of Montana.
What We Do
It starts with a statistical model—a mathematical representation of a system. We subconsciously use models all the time, like when we choose a line to stand in at the grocery store based on which one we think will move the fastest. The problem with using mental models is that they’re often wrong. After all, how many times have you gotten stuck in the slow grocery line? In wildlife biology, statistical models help take the guesswork out of management and can help clarify and defend decisions by providing a solid mathematical framework. We need models to help answer questions, evaluate outcomes, measure uncertainty, fix common data problems, overcome limited data, establish current population sizes and predict future population trends, just to name a few uses.
At Speedgoat, we build customized, server-based computer software that makes statistical models and analytics more approachable and helps bridge the gap between the biologists with mud on their boots and the computer geeks with three monitors each (that’s us). Our overall goal is to provide consistency and transparency for our collaborators, so that agency biologists can spend more time thinking about biology and less about spreadsheets.
We built our first software package, called PopR, in 2014. The first few versions ran on individual desktops, which became problematic when computers became outdated and often meant biologists in the same state were working with different sets of data.
To fix this, we moved to a server-based system and created an easy-to-use user interface. When our collaborators want to utilize the program, they log on to a simple website from any device with an internet connection and punch in what information they’re looking for and any additional parameters. The website reaches out to the server. The server identifies which model to run and what data is needed, securely connects to the agency database to collect only the required data, runs the model and spits back out a report to the user, all in usually under a minute. It works a lot like Google when you’re looking for a restaurant recommendation.
We also specialize in workflow management. Over the course of several visits, we sit down with our collaborators, from field techs to agency leadership, to discuss goals and the strengths and weaknesses of how data is currently collected and analyzed. Then we help design data collection strategies to best fit these goals as well as agency budgets. We also design systems to store the massive amounts of digital data collected by biologists and to polish that data until it shines. When the data is clean, the analysis comes easier. The way we see it, it’s all connected—a way to drive a data roundabout instead of having to turn left across incoming traffic and down a side street only to turn around, backtrack and do it all again for the next step in the process.
PopR started out as a research project, but the results had useful and concrete applications. And we started working with other collaborators to design new software packages, customized to different species and locales. Legally, we can’t house non-research projects under the umbrella of a research institute; those funds and enterprises have to be kept separate. To do just that, we created Speedgoat in 2016 using a technology transfer in cooperation with the Blackstone Launchpad at the University of Montana, a campus program which helps teach and develop entrepreneurship. Speedgoat comes from the nickname for pronghorn. The fastest mammal in North America, speedgoats are one of a kind—a little odd as the only member of the Antilocapridae family but a staple and symbol of the American West. We’re not nearly as fast as pronghorns, but our software is pretty speedy, and, as the place where biology, tech, and statistics intersect, we hope to fill a unique, but important niche all our own.