Leading Large Research Collaborations

Governance, financial stewardship, scientific renewal, and lessons learned from managing a multi-institution research center

Large research collaborations can achieve things that individual laboratories cannot. They can assemble complementary expertise, build communities, train scientists, share facilities and tackle problems whose scale exceeds any one institution.

But scale creates another set of problems. Scientific authority, financial responsibility and institutional accountability become distributed among people who may work hundreds or thousands of miles apart. Investigators have responsibilities to their own institutions as well as to the collaboration. Decisions that were once informal suddenly involve subawards, federal regulations, conflicts of interest, committees and multiple chains of authority.

My experience directing a large multi-institution research center convinced me that good science is not enough. Large collaborations require deliberately engineered governance.

Shared governance should establish direction and guardrails. It should not require collective voting on every operational decision.

1. Large collaborations are organizations, not just large grants.

A research center is simultaneously a scientific collaboration and an organization

A multi-institution research program may begin as a group of scientists with a common scientific objective, but once it acquires substantial funding, employees, subawards, educational programs and shared infrastructure, it becomes something more. It has many of the characteristics of a distributed organization.

That distinction matters. Scientific collaboration tends naturally toward discussion and consensus. Organizational management cannot operate entirely that way. People need to know who is accountable for decisions, who must be consulted, what requires formal approval and what authority has already been delegated.

Ambiguity may remain harmless while everyone agrees. It becomes dangerous when there is disagreement.

That was an important lesson from BioXFEL. The later UB/Research Foundation review found that the Steering Committee had been given an overall supervisory role, but that there was no additional governing framework or charter explaining how that responsibility should operate in practice. The reviewers specifically identified resulting ambiguity about which actions formally needed to come before the Steering Committee.

Governance documents matter least when everybody agrees—and most when they do not.

The BioXFEL center was organized around control of the Scientific Strategy by a steering committee having representatives from each institution involved. A finance and resource committee which was formed later in the center's history; the director and operations manager governed day-to-day execution, institutional compliance by the lead institution, and sponsor approval from the funder.

What happened in practice?

An independent UB/Research Foundation review subsequently concluded that the absence of a written governance framework created ambiguity over which matters had to be brought to the Steering Committee.

2. Define decision rights before the first disagreement

Collaborative science benefits enormously from discussion, criticism and collective judgment. But that does not mean that every administrative or financial action should require a vote.

A center should distinguish strategic governance from operational management.

The Steering Committee should establish scientific direction, consider major changes in resource allocation, evaluate Center-wide performance and advise on decisions that materially affect participating institutions.

The Director and management team should then have sufficient delegated authority to implement the agreed program. Routine expenditures, personnel administration and implementation of previously approved activities should not repeatedly return to a large committee unless a predefined threshold has been crossed.

There must also be a mechanism for exceptional decisions. Major reallocation of funds, termination or creation of substantial activities, significant changes in institutional commitments, conflicts involving senior leadership, or matters requiring sponsor approval should trigger additional review.

The important point is not where any particular threshold is placed. The important point is that the threshold is written down before somebody needs to invoke it.

Consensus is a valuable way to set direction. It is a poor mechanism for purchasing a box of laboratory supplies

 

An example is Director spending authority becoming contested; even a proposed expenditure of approximately $73 generated emails and votes, This illustrates how quickly governance can turn into transaction-level management.

3. Establish a Finance and Resource Committee on day one

One change I would now recommend for almost every large distributed research center is the creation of a Finance and Resource Committee at the beginning of the award—not in response to a dispute.

Its purpose should not be to run the Center or substitute itself for the Director or Steering Committee. Its purpose should be to make financial information routinely visible, understandable and independently reviewable.

The committee should review budget performance, major variances, institutional allocations, personnel commitments, subaward changes, projected carryover and proposed material reallocations. It should also provide a defined route for consideration of financial matters involving potential conflicts of interest.

Routine decisions already contained within an approved budget should remain delegated unless they cross agreed thresholds.

Financial transparency should be routine enough to be boring.

A recommended charter

This is where I would put your one practical checklist on the page:

Membership should include financial/grants expertise, appropriate institutional representation and enough independence that it cannot become an informal faction of the scientific leadership.
The committee should receive a regular dashboard containing approved budget, expenditures, commitments, projected year-end balance, major personnel effort, subaward status and material variances.
Dollar and percentage thresholds for additional review should be established in advance.
Recusal procedures should apply whenever a committee member's institution, laboratory or personal interests are directly affected.
Minutes should document decisions and reasons, without trying to recreate every discussion.
The committee advises or approves only those matters explicitly assigned to it; it should not gradually acquire undefined executive authority.
Formal financial compliance remains with the responsible institutions and awardee.

The assessment tested expenditure controls, segregation of duties, reimbursement, payroll/effort and other grant transactions and reported no exceptions in the financial controls it tested. Its conclusion stated that no instances of non-compliance were identified during the periods examined and that the original allegations were not identified as exceptions in its expenditure testing.

Strong financial oversight is valuable even when nothing improper has occurred, because it protects the Center as well as the people running it.

4. Transparency does not mean management by committee

A common reaction to concerns about governance is to demand that more decisions be brought to more people. That can increase visibility, but beyond a point it reduces accountability rather than improving it.

A system in which ten people nominally make every decision can become a system in which no individual is clearly responsible for any decision.

A better model combines high transparency with explicit delegation.

Budgets can be visible without being collectively administered. Decisions can be documented without requiring unanimous consent. Committees can question decisions without assuming executive authority. Directors can have operational discretion without having unlimited authority.

Good governance therefore asks two different questions:

Who should know about this?

and

Who has authority to decide it?

Those answers need not be the same.

5. Conflicts of interest must be managed, not merely disclosed

Large collaborations almost inevitably create overlapping roles. A Director may simultaneously be a professor, an institutional leader, an investigator and the representative of one participating institution. Steering Committee members may make decisions affecting funding for their own laboratories. Institutional administrators may simultaneously serve the interests of the Center and their employer.

Those overlaps do not necessarily imply improper conduct. They do, however, create actual or perceived conflicts that should be addressed prospectively.

I learned that simply filing a conflict disclosure is not enough. The organization needs a mechanism to determine whether the disclosed relationship requires recusal, independent review, monitoring or some other management plan.

That conclusion is supported directly by both independent reviews. Freed Maxick concluded that the relevant conflict had been disclosed but recommended active conflict management, including establishing a committee to evaluate potential conflicts, develop management plans and monitor compliance.

The UB/RF review similarly found that potential conflicts associated with overlapping institutional roles had been disclosed, but that adequate follow-up and management had not occurred.

A conflict disclosed but not managed remains a governance problem.

6. Scientific programs also need governance

Financial governance tends to receive attention because the rules are explicit. Scientific resource allocation can be harder.

Successful centers accumulate projects, themes, committees and commitments. Some continue to produce important science. Others reach their natural conclusion. Still others may have been excellent ideas when the proposal was written but become less compelling as technology, competition or scientific opportunity changes.

Stopping an activity is difficult because scientific programs acquire constituencies. People, students, collaborations and professional identities become associated with them.

For that reason, scientific review should be built into the organization rather than occurring only when resources become scarce.

Every major scientific activity should periodically be asked:

What question are we trying to answer now?

Has the field changed?

What have we learned?

What is the next decisive milestone?

Are we uniquely positioned to do this?

What are we not doing because we continue doing this?

The purpose is not to punish risky research. A center should be able to support projects that fail scientifically. The danger is supporting projects that no longer have a meaningful test of success.

A successful research center must be able to start promising ideas quickly—and stop yesterday's promising ideas gracefully.

Sunsetting is not synonymous with failure. Completion, technological displacement, changing scientific opportunity and diminishing marginal return are all legitimate reasons to redirect resources.

7. Separate scientific review from financial politics

One danger in a mature consortium is that institutional allocations gradually become treated as entitlements. The annual question changes from “What science should we do?” to “How much does each institution receive?”

That reverses the proper order.

The scientific portfolio should first be evaluated on the basis of the scientific opportunity and the Center's objectives. Resources should then follow the selected work.

Where possible, periodic scientific review should include people who do not directly benefit from the funding decision. An External Advisory Committee can be particularly valuable here.

Institutions should not own pieces of a research center. They should contribute to a common scientific mission.

8. Build a culture in which disagreement can occur safely

Large collaborations need disagreement. Scientists should be able to challenge scientific priorities, budgets and leadership decisions. But disagreement needs a process.

Concerns about financial compliance should be investigated through established institutional mechanisms. Scientific disagreements should be debated scientifically. Personnel issues should be handled through appropriate employment processes.

When those categories become mixed, disagreements can escalate rapidly and become personal.

One lesson I took from my experience is that allegations of financial impropriety are sufficiently serious that they should be handled professionally, independently and on the basis of evidence. That was also the position I recorded contemporaneously during the BioXFEL disput

Challenge decisions vigorously. Investigate serious allegations independently. Do not confuse either process with personal conflict.

9. The human side of leadership

eople do not leave their identities, families, values or personal circumstances behind when they become scientists or administrators. Leadership therefore also requires a culture in which disagreement over personal beliefs does not become professional retaliation or organizational dysfunction.

My own experience reinforced the importance of maintaining clear boundaries between personal values and professional responsibilities, protecting the privacy of family members, and insisting on respect even where individuals hold very different personal views.

A research organization should make expectations concerning nondiscrimination, professional behavior, confidentiality and respectful conduct explicit before a conflict occurs.

10. A documented case study

BioXFEL was a large NSF-supported Science and Technology Center involving multiple institutions and a distributed scientific and administrative structure. Like many successful collaborations, it developed governance practices while simultaneously conducting ambitious science.

In 2021, serious concerns were raised concerning financial decisions, transparency and the relationship between the Director, Steering Committee and participating institutions.

Those concerns triggered two substantive reviews.

This episode is useful not because other centers will experience exactly the same events, but because it demonstrates what can happen when questions of financial oversight, delegated authority, shared governance, conflict management and personal trust collide.

11. What the independent reviews actually found

The reviews did not conclude that every aspect of administration had been ideal. Neither did they substantiate the principal allegation that Center funds had improperly been repurposed.

The UB/RF review stated that the evidence did not support the allegation that previously approved BioXFEL allocations had been repurposed without Steering Committee knowledge. It further found no direct evidence that Center funds had been used outside their intended purpose or project scope. At the same time, it identified improvements needed in governance, communication, conflict management and post-award administration.

Freed Maxick's independently commissioned assessment likewise reported no instances of non-compliance in the transactions tested and stated that the allegations examined were not identified as exceptions in its expenditure testing. It nevertheless recommended stronger conflict-of-interest management.

That combination is, in my view, the most useful lesson from the entire episode:

An organization does not need wrongdoing before it needs better governance.

12. What I would put in place from day one

If I were starting another large research center

1 A written governance charter
Who decides, who advises, who must be informed.

2 A Finance & Resource Committee
Routine, independent visibility into financial activity.

3 Explicit delegation
Operational authority within agreed limits.

4 Predetermined escalation thresholds
Major changes receive additional review.

5 Active conflict-management plans
Disclosure alone is insufficient.

6 Annual scientific portfolio review
Continue • Expand • Pivot • Complete • Sunset.

7 Independent channels for serious concerns
Allegations are examined professionally rather than politically.

8 Documented decisions
Enough record to reconstruct why important decisions were made.

9 An empowered External Advisory Committee
Independent scientific perspective and periodic strategic challenge.

10 A professional-conduct framework
Scientific disagreement must not become personal warfare.

Design governance for the disagreement you hope never happens, not merely for the collaboration you have today.

13. Source documents

UB / Research Foundation Internal Audit memorandum - August 11, 2021 - Independent institutional review of the allegations, grant administration and governance structure.
Freed Maxick Grant Assessment - May 2021 - Independent assessment commissioned by HWI examining expenditures, effort, controls, governance and conflict-of-interest procedures.
HWI Board letter to the BioXFEL Steering Committee - May 5, 2021 - Contemporaneous institutional response from the HWI Board following receipt of the Freed Maxick assessment.
Edward Snell statement and background - May 2021 - The Director's contemporaneous account of the dispute, including correspondence and his interpretation of the events as they unfolded (a personal contemporaneous account and not an independent review).

14. A note on the documentary record

These materials are provided so that readers can distinguish contemporary claims from subsequent findings. The independent assessments, institutional correspondence and my own contemporaneous statement represent different perspectives and should be read as such. My purpose in publishing them is not to relitigate individual disputes but to show how governance weaknesses can amplify disagreement in a large scientific enterprise—and how future collaborations might be structured better.

15. Closing section

The scientific achievements of a large collaboration depend primarily on creative people, important questions and the resources to pursue them. But those ingredients alone are not sufficient.

A center must be capable of making decisions, reallocating resources, resolving disagreement and periodically questioning its own priorities.

Governance should be strong enough to protect public funds without becoming bureaucracy for its own sake. Leadership should have enough authority to act without becoming unaccountable. Scientific programs should receive enough stability to attempt difficult things without acquiring a permanent claim on resources. People should be able to challenge leadership without serious allegations becoming instruments of ordinary scientific disagreement.

The most important lesson I took from managing a large distributed research enterprise is therefore a simple one:

Good governance does not compete with good science. It creates the conditions in which good science can survive disagreement, change and success.