NOAA Fisheries' Office of Science and Technology (OST) and Office of the Chief Information Officer (OCIO) are designed to function as an integrated engine — OST providing scientific expertise, OCIO providing the technology infrastructure, together delivering solutions to the fisheries scientists, marine ecologists and enforcement officers who depend on them. By early 2024 the stakes were unusually high. Climate change was accelerating the pace at which marine ecosystems were changing, AI and machine learning were altering what was possible in fisheries science, and a $3.3 billion investment from the Inflation Reduction Act had put unprecedented — and temporary — funding into the agency. The clock was running to use it before it ran out.
The direction was not in dispute. Both offices agreed that NOAA Fisheries needed cloud infrastructure, open data practices and modern tooling, and both wanted to deliver it. What did not yet exist was an operating model that let them deliver it as one service rather than two.
In practice, users received conflicting platform guidance, and alignment between the two offices took longer than the work allowed. Scientists across the regional centers increasingly built their own solutions rather than routing requests through OST and OCIO. As one staff member described it: "It sometimes feels like we're bobbing along in our own individual life rafts, when we should be sailing together in the same boat." The IRA investment raised the cost of that gap: a once-in-a-generation opportunity to modernize had arrived, and the operating model for delivering on it together was not yet settled.
The gap was not invisible — people inside the system could describe it precisely. As one participant put it:
Becoming a unified source of forward-looking science and technology solutions was not one obstacle but six interlocking ones, each with its constraint owned by a different part of the system. None could be resolved by OST or OCIO acting alone. These six dimensions became the six topics the room was built around.
| Dimension | Why it could not be solved alone |
|---|---|
| Change management | Security and FISMA policy boundaries that make change slow are set and interpreted outside either office, and interpreted differently from site to site. Any change process that would actually hold had to be agreed with the regional fishery management centers who would have to run it, not designed in headquarters and issued. |
| Future mindset | Innovation depended on capacity that day-to-day operations consumed first. Headquarters was expected to think ahead while the regions ran reactively against immediate demands; neither could rebalance that on its own without the other absorbing the cost. |
| Leadership | Accountability for a joint vision spans two reporting lines, two sets of performance plans and two budgets. A shared standard for what leaders are answerable for cannot be declared by one office for both. |
| Unified operating model | Duplicate intakes, two project management offices and parallel cloud support are only visible as duplication from a vantage point above both offices — and any consolidation lands on the science centers that consume the services, who therefore had to be part of designing it. |
| User focus | The two offices could not agree who counted as a user — the public, the regions, other line offices, or each other. That question can only be settled with the users in the room, and a service catalogue written without them describes what is produced rather than what is needed. |
| Workforce & partnerships | Hiring timelines, contracting mechanisms and cooperative-institute rules sit with HR authorities and partner organizations beyond both offices. The levers that would change them are not held inside NOAA Fisheries at all. |
MMG convened 31 participants over three days at NOAA headquarters in Silver Spring, Maryland, on 26–28 March 2024: 12 from OST, 6 from OCIO, 8 scientists and IT staff from the regional science centers and regional offices, and 5 from partner offices — OAR, NESDIS and NCEI, and the Office of Habitat Conservation. A further group attended as observers. The composition was the design choice that mattered most: the people whose workarounds were the evidence of the problem were in the room while it was being solved.
| Stakeholder | Role in the problem | Why their absence would have stalled the solution |
|---|---|---|
| OST staff | Own the science, data and statistical products, and the requirements that any platform has to satisfy. | A technology roadmap written without the science requirements produces infrastructure that scientists route around — which is precisely the pattern the workshop existed to break. |
| OCIO staff | Own the infrastructure, the security posture, and the cloud and application services. | Every recommendation about shared platforms, self-service or open tooling depends on people who can say what is technically and legally possible, and who carry the risk when the answer is yes. |
| Scientists and IT staff from the regional science centers | Are the users. Their workarounds were the most accurate available diagnosis of where the two offices were failing to connect. | Both offices had already tried to fix this between themselves. A user-focus strategy built without users would have repeated the error at higher cost. |
| Partner line offices — OAR, NESDIS and NCEI | Face the same integration problem elsewhere in NOAA and hold enterprise capabilities that NOAA Fisheries would need to use rather than rebuild. | Several recommendations only make sense at NOAA scale. Designing them without the offices that would have to interoperate would have produced a plan that stopped at the agency boundary. |
| The Office of Habitat Conservation | Is an internal consumer of both offices' services from outside the science-center network. | A definition of "user" drawn only from the science centers would have been too narrow to survive contact with the rest of NOAA Fisheries. |
| The two office directors | Hold the authority to act on whatever the room produced. | Both directors worked the topics alongside everyone else, so the recommendations were built with their constraints already in view. A sponsor who watches the reasoning develop is far harder to surprise with a finished document. |
Evan Howell, then Director of OST and now Deputy Assistant Administrator for Scientific Programs at NOAA Fisheries, and Nancy Majower, then Chief Information Officer of NOAA Fisheries' OCIO and now NOAA's Chief Information Officer, had both invested in bilateral leadership meetings, internal working groups, and direct conversations about alignment. The obstacle was structural rather than personal: the two offices had different cultures, different incentive systems, and a long institutional history of operating independently. Every coordination attempt ran into the same limit — the conversation stayed within leadership, and the scientists who depend on both offices were not in the room when integration was being discussed. As one senior NOAA Fisheries leader put it:
What was missing was not another diagnosis. Both offices could already describe the problem accurately. What was missing was a mechanism for the people who would have to live with the answer to build it themselves, under a structure that forced every proposal to be tested by colleagues who would bear its cost.
Discovery interviews, planning sessions with both offices and background research were distilled into a single, deliberately time-boxed question that framed all three days:
Two things in that sentence did the work. The first is our. The question is not what one office should do differently; it commits both offices, in the same breath, to changing mindset, workforce and operations together — which makes it impossible to answer by assigning the problem to the other side of the house. The second is 6 months, defined as actions that could start, though not necessarily finish, inside that window. A six-month horizon rules out the comfortable answer — a multi-year roadmap nobody has to begin this quarter — and forces the room onto what can be started against current budgets, current staffing and current authority.
Going in, the sponsor was explicit that the measure of success was not the document:
Six topics were the instrument — Change Management, Future Mindset, Leadership, Unified Operating Model, User Focus, and Workforce & Partnerships — designed during discovery so the challenge would be opened along every seam at once: why change had not stuck, what a future-focused agency would look like, who is accountable for the vision, how the two offices should actually operate, who they serve, and who they need to become. Six teams worked those seams in parallel rather than in sequence, which is what made sixteen working hours across three days sufficient.
Each team is split by role. Members own the topic and are accountable for producing three recommendations that answer the Question through the lens of that topic. Critics sit in the same session with a standing mandate to attack the work and are invited to do so at fixed points. Challenge is therefore structural rather than a matter of who feels safe speaking up — and because every participant sat on more than one team, the reasoning behind each recommendation travelled across topics instead of staying with the group that produced it.
Each topic ran the same three-stage discipline. In Analyze, teams surfaced the real constraints. In Diverge, they generated options widely before committing to any. In Converge, each team built the three recommendations it would own, and the full room then voted to prioritize the resulting 18. What follows is how that played out, topic by topic. The mechanics were identical; the content differed.
Analyze. This team's brief was to work out why the organization was so resistant to change and what would make it better at undertaking change. The constraints it found were mostly structural. Government policies outlast their purpose and cost years of effort to revise; the same policy is interpreted differently at different sites, so staff get different answers to the same question; and the people who carry the risk of approving something carry none of the cost of refusing it, which makes no the safe answer. A single objection by email can stop an initiative outright. The team's own framing was precise: the binding problem was less opposition to change than operational inertia, in an IT function already unable to keep pace with existing workload.
Diverge. Options ranged from a national change-management appeal board, made fully transparent, so that an unworkable policy requirement has somewhere to go; to a single named point person accountable for change; to collapsing the number of FISMA system boundaries — considered even as a thought experiment, down to one — to see what it would unlock; to reframing shadow IT as innovation channels with sandboxes and a formal route to production; to town halls and roadshows that made the benefit of a change visible before it landed; to an explicit off-ramp for people who did not want to change; to training leadership to bring staff through a change rather than to it.
Converge. The team committed to three recommendations: create one change management process, led nationally and including membership from all regional fishery management centers, to improve oversight and eliminate fiefdoms; establish and manage a formal transition process from sandbox to operations, with retrospectives held whether something fails or succeeds; and facilitate change through better communication, meaning two-way channels, a feedback mechanism that guarantees a response, and materials that carry the vision rather than announce it. The sandbox-to-operations recommendation ranked fifth of the 18 in the room's vote.
Analyze. This team examined the current mindset and what stood in the way of a future-focused one. Grassroots innovation already existed and could not scale: individuals carved out space for it only when their role was not delivering day-to-day commitments, and there was no arena at headquarters to try things out. Risk aversion compounded it — science assumes you will not get it right the first time, and the surrounding system does not. The team also named a geography problem. Regional offices run reactively against immediate demands while headquarters is expected to think ahead, which risks headquarters dictating a future the regions have no bandwidth to build.
Diverge. Options included creating a dedicated innovation function sitting between the two offices; jointly funded positions and an agile workforce shared across them; an annual technology week bringing the OST and IT communities together; scrum teams pairing OCIO staff with scientists; hackathons; a mutual list of grievances that each side would have to produce a plan to address; centralizing GitHub across all centers; requiring that any new application be open source or carry a justification; a digital object identifier for every dataset; turning off a quarter of the application estate; sabbaticals and industry internships; and mandated focus days protected from meetings.
Converge. The team committed to: create an Office of Transition to identify and manage innovation projects, piloting the sequence from innovation to proving ground to evaluation to operations with in-year funds; realize enterprise solutions for open source, open science and open specification, with one location for accessing NOAA data and every dataset cited, linked and accessible; and manage the footprint — rationalize software, identify what to stop — to create the capacity that innovation requires. The Office of Transition ranked second of the 18 and the enterprise open-source recommendation third, the highest-placed pair from any single topic.
Analyze. This team worked on defining and empowering leaders at every level. Its analysis was the most candid in the room. Leadership had become a weakened word internally because follow-through had been inconsistent; transparency on budget, staffing and prioritization was thin; at senior levels ground truth is hard to reach and the echo chamber is loud; and there was no filter on what the offices were asked to absorb, which produced burnout rather than prioritization. The team also observed that most leaders had arrived through science or technical routes and had been promoted for that, not developed as leaders.
Diverge. Options included joint core values and a shared set of three to five goals; realigning the internal communications function toward the change rather than outward; a visual map of where the two offices' roles overlap and where the gaps sit; pushing risk tolerance down to the level where the work actually happens; collapsing layers so leadership sat closer to the ground; blameless post-mortems as standard practice; transparent performance plans; a joint OST/OCIO element written into both offices' plans; merging the two budgets; and picking one cross-cutting problem to solve publicly as a demonstration of what the two offices can do together.
Converge. The team committed to: enable the co-owned recommendations coming out of the other five topics — its brief was explicitly an enabling one; communicate a unified, actionable vision that shows every person their role in it; and ensure accountability on progress toward that vision, through cross-office teams, transparent performance plans, objectives and key results linking the groundwork to the vision, published progress, and top cover for the people taking the risks. Accountability ranked eighth of the 18. The two vision-and-communication recommendations placed last — a result worth reading plainly rather than explaining away: this room wanted mechanisms and follow-through, and had limited appetite for another statement of direction.
Analyze. This team worked on how the two offices could realize benefits from working together and remove the us-versus-them framing. It found duplication that only becomes visible from above both offices — two project intakes, two teams supporting the same cloud infrastructure — and no human integration plan behind any of it. Roles were not clearly defined. The word "unified" was being heard as loss rather than benefit. And underneath sat trust: the team was blunt that the offices' delivery reputation had not been high, and equally blunt about data sharing, which is a technical problem, an emotional one and a political one at once. Nobody, the team noted, had ever been penalized for refusing to share.
Diverge. Options included prioritizing by risk structure rather than organizational structure; establishing risk-tolerance zones so risk could be accepted in some areas and contained in others; removing FISMA boundaries between systems; a single ticketing system that increases transparency about who is fixing what; a published prioritization matrix so people can see what is being worked on; enterprise CI/CD on a central GitHub; a data-sharing pilot run specifically to demonstrate the benefit; a service catalogue and a skills inventory; radical transparency on budget, structure and people; and small empowered teams with real decision and risk authority.
Converge. The team committed to: create a new shared platform for Fisheries cloud services, organized by function and risk rather than by fishery management center, with a roadmap extending beyond infrastructure to publishing and GIS and with AI and machine learning toolkits integrated from the start; develop a centralized application intake system built on open source and open specification, with cross-functional teams supporting both platform and applications; and pilot enterprise self-serve data storage, beginning with environmental data. The shared cloud platform was the single highest-voted recommendation of all 18, taking 13 per cent of the room's votes.
Analyze. This team defined users broadly — the public, the regions, and staff inside OST and OCIO — and immediately hit the first constraint: the two offices could not agree who the user was. There was no service catalogue, so users could not find out what was available. From outside, the two offices looked similar enough that people could not tell where to go for what, and support was piecemeal. IT capacity was structurally smaller than user demand, and security implementation was frequently the reason a request was declined. The team also named a trust problem directly: users should not have to know someone inside the office to be heard.
Diverge. Options included defining distinct user groups and surveying their unmet needs; a user service catalogue; a single entry point combining a help desk for immediate support with an advisory service for people who need navigation rather than a ticket; self-service as the default model, on the basis that the user is often the expert; a standing rule to answer yes by default and document every no; users seated on decision-making bodies; a user ombudsperson for when everything else fails; shadowing exchanges between scientists and IT staff; a standard Fisheries-wide non-disclosure agreement to make sharing possible; and support for self-organizing user communities.
Converge. The team committed to: shift to a user-centric model, defining distinct user groups, building a service catalogue, and standing up a team that reviews user feedback regularly — including what gets declined and why; unify the support and innovation process across Fisheries, starting with OCIO and OST, to create a single entry point and reduce shadow IT; and establish a culture of sharing across code, data, processes and documentation, using open-science platforms and a Fisheries-wide NDA to make it possible. The culture-of-sharing recommendation ranked fourth of the 18 and the user-centric model sixth.
Analyze. This team worked on the roles the two offices need, the balance of skills and perspectives, and the partners required to close the gap. The constraints were concrete. There was no mechanism to matrix people across the two organizations, and blended-workforce confusion followed wherever roles and responsibilities were undefined. Reliance on contractors had become a standing answer to short-term capacity, with unclear boundaries between contractor, cooperative-institute and federal roles. Hiring moved slower than the change it was meant to support. And job postings asking for a degree "or applicable relevant experience" quietly screened out capable applicants who did not know how to translate their experience into that language. The team also noted that the offices often did not know who their own partners were.
Diverge. Options included universal focus days with no meetings, made consistent across Fisheries; cross-regional communities of practice; mentorship cohorts pairing one mentor from each office; an ability matrix to make matrixing possible; a joint OST/OCIO staffing plan exposing gaps and duplications; universal position descriptions and full use of the available hiring authorities; recruiting on the spot at hiring events; solution-based contracting rather than staff augmentation; a dedicated HR liaison rather than a collateral duty; a partner story map showing every project and the partners behind it; and rules of engagement with a menu of partners tied to project life cycles.
Converge. The team committed to: promote innovation through professional development, engagement and upskilling, beginning with universal focus days made consistent across Fisheries and formal mechanisms to report back what those opportunities produced; include partners earlier in the solutions development process, backed by an inventory, rules of engagement and a partner story map, rather than treating partners as an afterthought; and get and keep the right talent, through a joint staffing plan, a unified OST/OCIO hiring process, expanded hiring authorities and a structured onboarding programme. The professional development recommendation ranked seventh of the 18.
Across the six topics the room produced one coherent strategy rather than six wish-lists. The 18 recommendations sorted into three themes — how we work: converge on open, unified platforms for applications and data; how we innovate: empower innovation through user focus and capacity building; and how we change: lead transformational change through vision and accountability. When the room allocated its votes the top three all pointed the same way: create a new shared platform for Fisheries cloud services (13.0 per cent), create an Office of Transition to move innovation into operations (9.9), and realize enterprise solutions for open source, open science and open specification (9.4). Read together, they say something specific — before this organization can change its culture, it wants shared infrastructure that both offices run together.
Leadership left with 18 prioritized recommendations and three things that years of internal effort had not produced:
In an anonymous post-workshop survey completed by 24 of the 31 participants, 22 said they hoped action would be taken on the 18 recommendations. The same number said the format helped the group make significant progress in a relatively short time, and that they could share ideas without fear of judgment or criticism. Twenty of 24 supported the recommendations of the teams where they had sat as a Member.
Buy-in also showed up in what people volunteered to do next. One senior OST data leader described the personal commitment she left with:
For the sponsor, the shift in the room over three days went beyond the plan on paper:
Because the problem wasn't a lack of communication at the top — it was a structural misalignment between two offices with different cultures, different incentive systems, and a long institutional history of operating independently. Every coordination attempt ran into the same friction: the conversation stayed inside the same small group of senior leaders, and the scientists who depend on both offices day to day were not in the room. A process that kept the same people talking in the same configuration was never going to produce a different result.
In an anonymous post-workshop survey completed by 24 of the 31 participants, 22 agreed or strongly agreed that they hoped action would be taken on the 18 recommendations. The same number said the workshop format helped the group make significant progress in a relatively short time, and that they could share ideas without fear of judgment or criticism. Twenty of 24 supported the recommendations of the teams where they had sat as a Member. At that sample size the margin of error is roughly nine points, so these are directional figures rather than precise ones — but the direction is consistent across every question asked.
By designing the engagement so that both offices work on shared problems rather than negotiate positions. MMG structured the session across six interconnected topics — change management, future mindset, leadership, unified operating model, user focus, and workforce and partnerships — so the conversation was always about what the integrated system needed, not about which office owned which decision. Each topic team was split into Members, who owned the topic and were accountable for producing three recommendations, and Critics, who sat in the same session with a standing mandate to attack the work. Challenge was structural rather than a matter of who felt safe speaking up. Scientists from the regional centers were in the room throughout, which kept the conversation grounded in operational reality.
The workshop produces a prioritized strategy with shared ownership — not an implementation plan with assigned owners and timelines. A follow-on session to assign owners, analyze roadblocks, and build out the execution roadmap is a natural next step, and it was part of what OST and OCIO committed to as part of this engagement. MMG can facilitate that session or advise on the design. The most important variable is sustained leadership attention on the commitments through the first 90 days.
If two parts of your organization are each doing their job well and still cannot deliver as one service — where the direction is agreed, the goodwill exists, and the operating model does not follow — that is the conversation worth having. Get in touch.