The WarGames Problem: Why Data Is Harder Than It Looks

How fictional computers set impossible expectations for real institutions

I saw WarGames in the theater in 1983. A few years later, I watched it again on VHS. I haven’t seen it since, but one image never left me. A teenager in his bedroom, a phone line, a few keystrokes, and the entire NORAD defense system was on his screen. The movie wasn’t asking you to believe that was realistic. But it, and dozens of films like it, quietly shaped what a generation came to expect from computers and from the information they were supposed to make available. Data was accessible. Systems talked to each other. The right answer was always a few clicks away.

That assumption followed us into our professional lives. It shows up in grant applications, strategic plans, and meeting agendas, where data sharing appears as the obvious first step, the easy item to clear before the group moves on to the harder work. What none of those movies ever showed us was the years of back-and-forth collaboration that had to occur before the first data exchange could happen.

The Meeting That Never Got Past Item One

Earlier in my career, I had the opportunity to observe a meeting that I have thought about many times since. I was working for the president of the university system in North Carolina, and she had convened a remarkable group of state-level leaders. Legislators were in the room. So was the president of the community college system, the secretary of labor, and a representative from the governor’s office, among others. The goal of building a stronger framework for collaboration across systems to produce a better-prepared workforce was ambitious, and the agenda they brought to the table reflected that ambition.

Data sharing appeared on that agenda as a preliminary item that everyone seemed to regard as perfunctory. The expectation was to dispatch it efficiently and move on to the harder work of designing programs and aligning systems. These were not naive people. They were among the most experienced and capable leaders in the state. And they spent the entire meeting on that one item before time ran out. The group never reconvened.

What I watched that day was not a failure of leadership or ambition. It was the WarGames assumption meeting reality. If a teenager in his bedroom could pull up NORAD in 1983, sharing data across a handful of state agencies should be straightforward. Except it never is. And the people who know that best are rarely the ones setting the agenda.

What the Work Actually Requires

Data sharing sounds like an information problem. In practice, it is an organizational problem, and the information piece is often the last thing you get to. Before two organizations can share data meaningfully, they must agree on definitions. What counts as a student? What counts as a completion? What counts as a successful outcome? Those questions sound simple until you discover that two agencies have been answering them differently for twenty years and have built their entire reporting infrastructure around those different answers.

Then there are the systems themselves, often built in different eras, by different vendors, for different purposes, with no expectation that they would ever need to speak to each other. And underneath all of it are the cultures of the organizations involved, the unspoken assumptions about what data is for, who owns it, who has the right to see it, and what it means to share it across an institutional boundary.

None of this is insurmountable. But none of it is perfunctory either. The distance between the dashboard someone imagines in a planning meeting and the one that eventually gets built is where initiatives stall, partnerships dissolve, and leaders quietly conclude that data-driven work is more slogan than practice.

Two Years Before the First Data Exchange

When I arrived in Harford County, I knew that realizing the vision I was building for students would require a genuine partnership with the local community college. Dual enrollment was central to that vision, and dual enrollment done well means that both institutions can clearly see the same students, track their progress in real time, and respond when something goes wrong.

I also knew, from what I had witnessed in North Carolina, that the data sharing piece was going to be harder than it looked. I discussed this with the college president when we began our conversations. I recalled that state-level meeting. I told my own staff the same thing before they began the detailed work of building the exchange. I wanted everyone who touched this effort to understand that difficulty was not a sign of failure. It was the nature of the work, and it took much longer than we had expected.

We spent two years on extensive back-and-forth between our teams before we reached a point where we could share enrollment and progress data daily. There were moments when my patience wore thin. But I understood what was happening beneath the surface. The friction was not about will or competence on either side. The cultures of a public school system and a community college are genuinely different, and those differences shape everything, including how data is structured, stored, labeled, and understood. In many respects, data sharing served as a proxy for the structural and cultural gaps the partnership had to close. Working through it built something neither organization could have built alone.

Seven years later, I was meeting monthly with the college president, by then the second person to hold that role since I arrived, to make sure both organizations kept moving in the same direction. We talked often about how fortunate we were. We knew colleagues in other communities who had tried to build similar partnerships and had never gotten traction. We also knew better than to declare victory. A partnership of that kind does not reach a finish line. It requires maintenance, attention, and leadership on both sides that never stops treating it as a priority.

When the Problem Is Inside Your Own House

The WarGames problem does not only appear at the boundaries between organizations. It lives inside a single school system, across departments that share the same building and answer to the same superintendent.

A few years ago, I began a close examination of special education performance and finance in Harford County. I wanted to build an internal dashboard that would allow us to identify patterns in student outcomes by school, disability designation, grade level, and other variables. I expected to find gaps in our data. What I found instead was that the data existed, it was extensive, and it was essentially unusable for the purpose I had in mind.

The problem came down to a reporting quirk that sounded minor until you tried to work around it. General education performance data was built around a snapshot taken on one specific date during the school year. Special education data was built around a different snapshot date. The two sources were difficult to align. Reconciling them took months.

I tell that story not as a cautionary tale about data systems, but as an honest account of what it means to work with institutional data in the real world. These were not poorly maintained systems. The people responsible for them were skilled and diligent. The problem was structural, built into the way reporting had been designed long before anyone imagined the question I was trying to answer.

What Leaders Owe the People Closest to Students

We ask teachers and principals to be data-driven. That expectation is reasonable. The decisions made in classrooms and schools every day benefit from good information, and leaders are right to build cultures where data informs practice rather than replaces judgment.

But the systems that underpin that expectation are genuinely hard to build and maintain, and the people who set it are often the furthest from the difficulty of meeting it. We ask teachers to use data they frequently cannot find, cannot access in time to act on, or cannot interpret without context they were never given. We ask principals to lead data-driven schools while managing the same infrastructure limitations their teachers face, plus a dozen others.

One important thing a district can do for its schools is remove that burden. Not simply tell teachers to use data better, but actually build the conditions in which useful, timely, actionable data reaches the educators who need it, without requiring them to spend the preparation time they do not have searching for it, compiling it, or waiting for it to arrive.

That is harder than it sounds. It took us two years just to share data with a college across town. But it is worth the effort, precisely because the people on the other end of that work are making decisions every day that affect students who cannot afford to wait.

 

Dr. Sean Bulson is a practitioner-scholar, professor of educational leadership, and the 2024 Maryland Superintendent of the Year. He served as superintendent of Harford County Public Schools from 2018 to 2026 and Wilson County Schools from 2011 to 2016. Subscribe for weekly posts on educational leadership, equity, and district transformation.