Simon Christopher Barry, 12 February 1965–16 July 2023

Abstract

Simon Barry was a statistical scientist of the highest calibre, a champion for the discipline throughout his illustrious career in government, the CSIRO and academia. He was a giant physically (a 6′ 5′′ frame with his frizzy hair seeming to gift additional height) and intellectually, who made a strong and lasting impression on all who encountered him. Tragically, Simon Barry died in a car accident on 16 July 2023, aged 58, leaving the world a diminished place.

Simon was born in Brisbane but grew up in Canberra, attending Pearce Primary School, Lyneham Primary School, Lyneham High School and Dickson College. He commenced an agriculture degree at the University of Sydney but transferred to Australian National University (ANU) after his second year where he studied botany. He started his honours degree working on the genetics of Onychophora a.k.a. peripatus, before switching to statistics, and graduating with first class honours in 1990. Field work in his thesis involved breaking open rotting logs to find peripatus, and then collecting them, along with any funnel webs also present in the logs (for a colleague doing a similar study).

Simon's first job was at the Australian Bureau of Statistics (ABS) and it gave him a valuable grounding in survey sampling and survey inference, but whetted his appetite for more. He subsequently joined the Australian Defence Force Academy (ADFA) in Canberra, working with Ted Catchpole and Ted's UK collaborators Byron Morgan and Steve Brooks on capture-recapture methods. While at ADFA, he commenced a PhD at the ANU on modelling truncated data (supervised by Terry O'Neill) and was awarded his PhD in 1996 (Barry 1995). His thesis received the P.A.P. Moran Prize at the ANU for its contribution to the Advancement of Probability or Statistics in 1999.

Simon joined the ANU as a consultant in the Statistical Consulting Unit with Ross Cunningham and Christine Donnelly and later as a lecturer in the then Department of Statistics and Econometrics. He collaborated with many in the ANU, but particularly those with a passion for ecology (Gibbons et al. 2000; Cunningham et al. 2006; Manning et al. 2006).

In 1999, Simon joined the Bureau of Rural Sciences (BRS), then the science research division in the Commonwealth Department of Agriculture, Fisheries and Forestry (DAFF). For the first few years he was the only statistician working in a mostly GIS group, but as Simon preached how statistics could change the lives of all the people in DAFF and demonstrated how he could do so, the team grew and he flourished personally.

One aspect that his BRS work provided was the opportunity to ‘fight fires’, as he put it. These were the real-world, big-impact projects where management decisions had the potential for real impact. Never shy of engaging in healthy dispute, Simon pushed to make changes in very significant areas such as import risk assessment, where Simon advocated for the need for rigorous and transparent quantitative approaches supporting what were then almost entirely qualitative processes with huge implications for trade. Simon worked on risk assessments for bananas, pigs, chickens and more. One memorable assessment related to apples, and the decision making behind why, for 90 years, Australia would not let New Zealand apples into the country. It was under constant scrutiny while Simon was in BRS, ending up in Senate estimates time and time again and ultimately in the International Court. DAFF relied heavily on Simon's expertise and ability to explain technical issues in an accessible manner and he was regarded very highly for the advances he made on such matters.

BRS gave Simon not only the opportunity to be close to the ‘action’ but also access to tools to innovate. At a time when very few Government organisations were utilising analytics beyond Excel capabilities, BRS had one of Australia's first Cray computers, enabling Simon to explore highly computationally intensive strategies. This led to important work in the early 2000s on the National Scale Land Use of Australia map (Stewart et al. 2002), which is widely acknowledged as a key source of statistical information for Australia and is in routine use.

Simon spent many years working in risk assessment for introduced marine pest species, collaborating with colleagues in CSIRO (Hayes & Barry 2008; Hayes, Inglis, & Barry 2019). This work had major influence, informing changes to both surveillance of marine pests and the regulations on shipping traffic aimed at minimising marine pest incursions and translocations in Australia. Some of Simon's work within DAFF contributed to the Intergovernmental Panel on Climate Change (IPCC), and saw him, as part of a large global team, awarded the 2007 Nobel Peace Prize.

In 2007, Simon joined the Mathematical and Information Sciences (CMIS) Division of CSIRO, where he became leader of a team focussed on environmental problems. In many ways it was natural extension of his work with the BRS, and CSIRO provided the deep collaborative and multi-disciplinary way of tackling important problems that energised him. He loved the contest of ideas, playing in other people's backyards, and strategising about the best way to tackle apparently unsurmountable problems at hand. Given Simon's leadership skills as well as his strong intellect background, it was no surprise that he progressed rapidly within CSIRO. For much of his time he led a large program of statistical and other analytics researchers and was a key member of various executive teams, working closely with Divisional Chiefs such as Bronwyn Harch and Louise Ryan. Simon was a senior leader in CSIRO's Data61 from near its inception in mid-2016, and acted as Chief of Data61 from September 2019 to July 2020. In his last role, he was Digital Lead for the Environment Business Unit within CSIRO.

While there were challenges stemming from changes to strategy and reorganisations to overcome, Simon thrived, transcending disciplines and administrative boundaries to forge many meaningful connections and collaborations. Simon thought deeply about the world—the natural world in particular—and was always interested in how that world could be modelled and represented in order to support better decision making. Sound statistical thinking, inference and logic were foundational to all his work, as he applied his skills in biosecurity, water resources, agriculture, climate change, ecology, marine, risk and other areas.

Despite the often heavy administrative load associated with leadership roles, Simon remained active as a researcher himself. For example, he played a crucial role in the methodology for the Bioregional Assessment Program that assessed the impacts of large coal mines and coal seam gas on water resources and water-dependent assets in Eastern Australia (Henderson et al. 2018); he worked closely with CSIRO colleagues to harness the digital revolution for agriculture (Barry et al. 2017); he evaluated trends in waterbird abundance (Caley et al. 2021); and he continued to assess biosecurity risk (e.g. establishing risk for invasive species at ports relevant to honey bees, Clifford et al. 2011; Caley, Paini, & Barry 2016).

Simon's technical contributions were often deeply rooted in statistical science. He believed in the primacy of data and its ability to shed light on the world. But he also knew how carefully that data had to collected and treated because it was often fragile, fragmented and biased.

He had a deep interest in statistical inference and the challenges of applying statistics to difficult problems. He was interested in problems with incomplete data where progress requires the collection of additional data and/or strong assumptions. Simon also enjoyed practically motivated theoretical challenges, especially if there was a provocative angle he wanted to push: for example, his work on truncation (O'Neill & Barry 1995a, 1995b), capture-recapture (Barry et al. 2003), zero-inflation (Barry & Welsh 2002), distance sampling (Barry & Welsh 2001), uncertainty quantification and info-gap decision theory (Hayes et al. 2013) and presence-only data (Ward et al. 2009) fall into this category. He thought a lot about the structured ways to incorporate expert opinions into a quantitative statistical framework for analyses and risk assessments in data-poor situations. His Point of Truth Calibration (POTCAL) methodology (Brookes et al. 2017) is arguably still a gold standard.

Simon maintained a strong interest in species distribution and abundance models (Barry & Elith 2006), including the establishment and dynamics of invasive species (Hayes & Barry 2008; Caley, Ramsey, & Barry 2015; Caley, Paini, & Barry 2016) and the potential role of citizen science and general surveillance in detection well ahead of the pack (Caley & Barry 2022). Methodologies for risk assessment in different contexts were always front and centre (Hood, Barry, & Martin 2009).

Simon always had a strong interest in computation. His excellent computational and coding skills were exemplified in his writing and running code to create maps for the 2003 Eureka-Prize-winning New Atlas of Australian Birds for Birdlife Australia (Barrett et al. 2003). These interests also meant that he was fascinated by emerging digital technologies, especially machine learning and artificial intelligence. The interface between those areas and statistical science was front and centre in CSIRO Data61, and while he was involved in many healthy debates about the relative merits or genesis of many different approaches, he was also instrumental in helping people to appreciate different perspectives, and ultimately to work together more effectively.

Simon was the architect of a cross-CSIRO initiative (the Machine Learning and Artificial Intelligence Future Science Platform) that sought to develop machine learning and artificial intelligence science relevant to CSIRO priority areas. While not reflected directly in the title, statistical methods and logic were heavily embedded. The platform involved widespread collaboration across CSIRO to make sure it attacked the right problems, and ultimately recruited 43 postdoctoral fellows as part of a step change in that activity area. The impacts and digital uplift from that 5-year initiative continue rippling through the organisation.

In the week prior to his death Simon spoke at the MODSIM 2023 conference in Darwin on ‘Modelling and the rise of machine learning’. He shared many perspectives in that talk, including his caution with the synthetic data movement and its proposal to generate data artificially as a solution to privacy and the high cost of collecting real data. He was concerned about the meaning and value of any subsequent inferences and felt that these developments could have major implications for modelling, science and policy development.

Simon wrote extensively throughout his career. He had over 7300 citations but inspired many more in others through his generosity with ideas, and left many ‘runs in the shed’ as some would say. Publishing was often not his priority: once a problem was solved to his satisfaction, he would often move onto the next problem.

It was in conversation though that he flourished and how many statisticians will remember him. Simon was someone who was engaged, generous, authentic, insightful, provocative, full of stories and colourful anecdotes. His ability to read the room and the problem, and come back with exactly what needed to be done in a way that got people engaged, was seen time and time again.

His legacy endures in those areas where he has promoted statistical thinking and logic, and compelled people to think deeply about the problem in hand and to match method appropriately with problem. It also endures in the people he developed and inspired through his generosity and way of seeing the world. When Simon left BRS, he left behind a large, stand-alone quantitative unit he had built (Information and Risk Sciences) and an important legacy that he imparted to many of his staff about the value of understanding the context of a problem and working closely with decision makers to get the best management outcomes.

Within CSIRO he was the most senior champion of statistical science, and managed to get it into all sorts of areas of CSIRO interest, and he developed and encouraged statistical disciples to take it further. He thought about the development of others through those individual conversations but also through our professional societies; for instance, he was President of the Australia and New Zealand Chapter of the Society for Risk Analysis in 2010 and 2011.

Simon was a formidable operator. His formal training in probability and statistics, sampling design and inference, computing, the multitude of relationships across government, university, and private sectors he cultivated, and his deep interest in the application domains and the people who inhabited them ensured that. We have all lost a valued friend and colleague who was truly committed to statistics.