The Court of Federal Claims has vacated the Army Corps of Engineers' roughly $28 million award to McMillen, Inc. for gate work at Webbers Falls Lock and Dam 16 in Gore, Oklahoma, ruling that the Tulsa District never explained why it gave McMillen a higher rating than Massman Construction Co. on a subfactor where the two companies wrote essentially the same thing. Judge Carolyn N. Lerner reissued the public version of her opinion and order in Massman Construction Co. v. United States, No. 25-2158, on July 7, 2026, after filing it under seal on June 15. The Corps is enjoined from starting performance and has 60 days on remand to fix the evaluation, cancel the solicitation, or take other action consistent with the ruling.
Background
The Tulsa District issued RFP No. W912BV25RA018 on July 11, 2025 — a FAR Part 15 negotiated procurement for the Webbers Falls Tainter Gate Replacement and Repair project. The final solicitation required the winner to replace or repair Gates 7, 8, 11, 5, 10 and 12 at the lock and dam, a component of the McClellan-Kerr Arkansas River Navigation System, a 445-mile waterway running from the Tulsa Port of Catoosa to the Mississippi River. On September 12, 2025, Amendment No. 5 moved replacement of Gate 5 out of the base scope and into Option 1.
According to the Corps' own announcement on DVIDS, the district awarded McMillen Inc. of Boise, Idaho roughly $28 million on September 30, 2025 — the district's final contract of fiscal 2025. The scope described publicly covered replacement of Tainter Gate 7 and repairs to Gates 8 and 11.
Massman protested. The agency asked the court to remand so it could take partial corrective action, and the court obliged in March 2026. The fix was narrow: USACE would issue a new contract exercising no options at the time of award. It did exactly that — and re-awarded to McMillen anyway. Massman came back to the court, and this time the government lost.
Key Details
Most of Massman's case failed. The court found the Corps had a rational basis for its Past Performance ratings and for its evaluation under the Technical Approach subfactor. It also held that the partial corrective action cured the alleged unlawful exercise of options at the time of award, and dismissed that count as moot. On the bulk of the protest grounds, the agency's judgment survived.
The award came apart on a single subfactor: Organization, under Factor 2, Management Approach.
Both offerors described a three-phase control process, and both used the same three phases — preparatory, initial inspection, and follow-up inspection. The court found the Agency had treated Massman's proposal differently when it was "substantively indistinguishable" from McMillen's. The Corps rated McMillen higher anyway and never rationally explained why. Lerner cited the standard from AshBritt: "[W]here an evaluation scores two virtually identical proposals differently, such an inconsistent evaluation is quintessentially arbitrary, capricious and unfair." As she put it, the court must defer to the agency's technical judgment, "but it cannot defer to judgment that is not there."
Organization was the only place McMillen came out ahead. Massman rated higher on Factor 3, Performance Capability. The two advantages offset each other, and that offset is what allowed the source selection authority to conclude the proposals were technically equivalent. Once the SSA found equivalence, price decided the competition — McMillen was lower — and no price-technical tradeoff was performed. The court rejected Massman's standalone attack on that tradeoff analysis; handling two genuinely equivalent proposals that way was defensible. The equivalence finding itself is where the case turned. The technical factors combined were significantly more important than price, so a tradeoff mattered — and the Corps never reached one.
Pull the Organization disparity out and the equivalence finding collapses, leaving the Corps holding a lower-priced proposal that, on this record, is technically inferior in a competition where technical merit dominated. The court framed the remand in exactly those terms: the agency "must either justify this disparity or explain why McMillen's lower-priced, but technically inferior proposal provides the best value to the government."
The relief tracks that reasoning. The court vacated the award decision, enjoined USACE from commencing performance, and remanded for 60 days to reevaluate, cancel the solicitation, or otherwise act consistent with the opinion. Massman's request for bid and proposal costs was denied as premature — the remand may yet hand Massman the contract.
What It Means for Contractors
The Corps evaluated defensibly almost everywhere, and lost on one subfactor — enough to stop a $28 million construction contract cold and put the Tulsa District back at square one more than nine months after award.
For protesters, that shape matters. Disparate treatment claims carry weight at the Court of Federal Claims precisely because they do not require the court to second-guess technical judgment. It need not decide whether a three-phase control plan is any good — only whether the record explains the different scores. When the answer is no, the AshBritt standard does the rest. Massman won by finding one place where the paper did not match the ratings — and by showing that the place it found was load-bearing.
That second half separates a sustained protest from a wasted filing fee. An unexplained rating on a subfactor that changes nothing gets a shrug. This one propped up the technical-equivalence finding, which eliminated the tradeoff, which handed the award to the lower price. When you debrief, trace the ratings forward into the source selection decision and ask which specific rating, if reversed, would have forced a different outcome. Protest that one.
Losing offerors should also read the corrective-action sequence as a warning. The Corps took corrective action on remand, narrowed it to the option-exercise issue, and re-awarded to the same firm. Partial corrective action buys an agency a dismissal on the count it addresses — Count III went out as moot here — while leaving the untouched grounds fully live. If your protest gets remanded and the agency's fix does not reach your core evaluation argument, that argument is still there when the award comes back.
There is a proposal-writing angle too. Both firms fell back on the same textbook three-phase vocabulary, leaving evaluators nothing to work with. Boilerplate gives an agency room to score you either way and defend neither.
For agencies, the message is narrower: the narrative has to carry the rating. The court was explicit that the APA requires the government to show its work. That is not a documentation quibble — it is the difference between a defensible award and a vacated one.
