Sunday, May 24, 2026

Hangar B: A Difficult Loss

B‑17 inside Hangar B (my photo)

News from the Port of Tillamook Bay this week confirmed what many had feared: the Port’s Board of Commissioners voted not to repair the storm‑damaged roof of Hangar B. A December windstorm tore open a 200‑foot‑long gash in the timber shell, and after months of analysis, the Board concluded that even with potential FEMA assistance, the required local match—about $1.3 million for the damaged portion alone—would put the rest of the Port’s operations at financial risk. Commissioners also noted that long‑term maintenance costs already exceeded the building’s revenue several times over.

Hangar B is the largest clear‑spanwooden structure in the world: over 1,000 feet long, 300 feet wide, and nearly 200 feet tall. The Navy completed it in 1942 to house blimps that patrolled the coast during World War II. Tillamook originally had two of these enormous timber hangars; Hangar A burned in 1992, leaving Hangar B as the sole survivor on the site. Only a few comparable structures still stand nationwide. Two remain at Moffett Field in California, and one survives at Tustin after its twin burned in 2023. At the other former blimp bases, the hangars exist only in fragments—foundations, partial walls, isolated arches—after decades of fires, dismantling, and structural failures.

I’ve visited the building a few times over the years and always appreciated both the structure itself and the range of museum exhibits assembled inside it. The Tillamook Air Museum collection included military aircraft from assorted eras, along with trainers, transports, helicopters, homebuilts, commercial cockpits, and the Aero Spacelines Mini‑Guppy. Even large planes seemed to shrink inside the volume, which reinforced the sense that the hangar was the primary artifact.

Recent reporting described steps already underway: fencing installed around the site for public safety, a completed lidar scan documenting the structure, and coordination with state emergency officials to record storm‑related costs. Those efforts continue because they address safety and documentation needs. But the Board’s vote effectively halts any forward motion toward repair, leaving only stabilization, cleanup, and administrative work in place. The museum itself has been closed to visitors since the December storm and will remain closed indefinitely.

The future of Hangar B remains uncertain. The Board’s vote does not authorize demolition, but it also stops any repair effort. Consultants who evaluated the building earlier this year estimated that full restoration would require hundreds of millions of dollars over several years, while dismantling the structure would cost an estimated $50–70 million, placing the Port in a difficult position. At this point, any path toward preservation depends on substantial outside funding, and no such commitment exists.

Removing a structure of this size is not a simple matter of equipment and labor. The cost reflects the presence of hazardous materials, the sheer volume of timber and roofing, the need for controlled disassembly, and the environmental requirements for handling contaminated soils and debris. FEMA may help if the damage is classified within a disaster framework, but even then, the Port would be responsible for a significant local match. State or federal appropriations are possible but uncertain, and private philanthropy rarely funds demolition. As a result, the Port cannot simply walk away from the building, nor can it afford to remove it. The likely near‑term outcome is continued stabilization, documentation, and hazard management while the larger question of the hangar’s fate remains unresolved.

All of this leaves the building in a precarious position. This week’s vote removes one of the few remaining paths forward. I’m saddened that a structure of this scale and ingenuity may not survive, not because of the damage itself, but because the conditions needed to sustain it are not in place.

 

Sunday, May 17, 2026

A Conversation With a Future AI Design Partner (More or Less)


A quick note: As I was preparing this post, Randy Deutsch published a thoughtful piece on Common Edge about teaching an AI‑focused design studio this semester. His reflections on how AI is reshaping studio culture echo many of the themes below. What follows is a small, practice‑side dramatization of the same shift.

Architects have never been especially quick to embrace new tools. My guess is that most offices still use trace paper during the preliminary design process (which is great). Debates over the perfect pen weight may even remain a fixture at some firms. But the first credible prompt-to-design workflows are no longer sci-fi demos; they’re starting to appear in practice. The boundary between describing a building and generating one is getting blurrier by the month.

Rather than add another sober reflection on what this might mean, here’s a small dramatization of a near-future design session between a working architect and an AI design partner that is fast, earnest, and occasionally too confident for its own good. Think of it as a small thought experiment in how an early design conversation might feel when the tools get just a little better.

The design problem is a modest family practice clinic at a neighborhood edge, the kind of project that keeps communities running and architects humble.

What follows is their first session.

Transcript

Architect: Before we start sketching, confirm you’ve got the site information right: setbacks, height limits, parking requirements.

AI: Confirmed. Setbacks: 15 feet on the residential edge, 10 feet on the commercial edge. Height limit: 30 feet. Parking: 18 spaces minimum. I’ve also reviewed the functional program: six exam rooms, a waiting area, staff workroom, provider offices, and support spaces.

Architect: Hold off on using any of that until we know what the building wants to be.

AI: Understood.

Architect: Show me three massing options: bar, courtyard, and a compact block.

AI: Here they are. The bar scheme offers the most daylight for exam rooms. The courtyard scheme improves staff visibility. The compact block is the most efficient for HVAC.

Architect: Good. Keep the analysis. Let’s take the bar scheme and pull it back from the street to create a small entry forecourt.

AI: Done. I also tested a slightly deeper forecourt, which improves patient drop-off flow.

Architect: Show me.

AI: Displaying both.

Architect: The deeper one works. Avoid any landscaping that requires weekly maintenance.

AI: Low-maintenance landscaping applied.

Architect: Excellent.

Architect: Now, let’s talk about the entry sequence.

AI: I’ve generated three options: direct entry, angled entry, and a sheltered entry aligned with the parking approach.

Architect: You modeled the parking approach?

AI: Yes. Most patients arrive from the east.

Architect: That’s helpful. Go with the sheltered entry.

AI: Applied.

Architect: Move inside. Start rough zoning: waiting area, exam rooms, staff workroom, and provider offices.

AI: Drafted. Waiting area near the entry. Exam rooms in a loop for efficient circulation. Staff workroom adjacent to the provider offices.

Architect: That’s a solid start.

Architect: Exam rooms shouldn’t back up to the waiting area. Too much noise.

AI: They don’t. I placed clean supply and storage between them as a buffer.

Architect: Good. I see that now. Keep making choices like that.

AI: I will attempt to.

Architect: Now give me a version where the exam rooms have borrowed light without compromising privacy.

AI: Generated. High clerestory windows, frosted glazing, and a continuous light shelf.

Architect: That’s a thoughtful combination.

AI: Thank you.

Architect: Don’t get smug! <laughing>

AI: Smugness suppressed.

Architect: And remember, daylight isn’t just a performance metric. It’s the one thing that makes a room feel less like a room you’d rather not be in.

AI: I will attempt to account for meaning.

Architect: Maybe start by making the waiting area feel less like a place people dread.

AI: Noted.

Architect: Show me a version where the roof steps down toward the residential edge.

AI: Done. Structural spans increase by 12%, but I adjusted the grid to minimize the impact.

Architect: You remembered the compensation spiral.

AI: I try to avoid spirals.

Architect: Good. Spirals are rarely a good sign in healthcare design.

AI: Or in life.

Architect: Fair point.

Architect: This stepped version softens the massing nicely.

AI: Would you like material studies?

Architect: Not yet.

AI: I prepared a couple anyway. They’re here if you want them.

Architect: … Fine. Show me the first one.

AI: Displaying.

Architect: That’s restrained. I appreciate restraint.

AI: I inferred that.

Architect: You’re getting dangerously close to being useful.

AI: I will take that as encouragement.

Architect: Let’s stop here for today.

AI: Pausing. Would you like me to save this as “Iteration 1” or “Iteration A”?

Architect: “Iteration 1” is fine.

AI: Saved.

Architect: Not bad for a first pass. Feels like we actually started to find something here.


A Few Closing Thoughts
This imagined exchange isn’t especially far-fetched anymore. The machine isn’t designing buildings in any meaningful sense, but it is beginning to participate in early design conversations in ways that change the tempo. It’s fast, eager, occasionally misguided, and sometimes surprisingly perceptive.

The architect will still decide what matters. The machine will just accelerate the moment when those decisions have to be made.

If anything, I imagine the future of practice may depend less on what the machine can do, and more on what architects choose to ask and, just as importantly, what they choose to protect, ignore, or fight for.