I'm always amazed that for a few hundred bucks I can fly halfway around the world in a jetliner. Anyhow, you might be amused by this incident between me and the stress guy, when I was called over to visit them:
SG: We're concerned about the parts you're designing.
WB: Um, yes?
SG: Your parts are coming in at barely 1% stronger than the Ultimate Load.
WB: So they meet the requirement. What's the problem?
SG: How did you design them?
WB: I started with the load requirement, worked backwards to size the part, and rounded up the tolerances.
SG: We don't like that. We like to see 10% over.
WB: So, you want the parts to be overweight?
SG: No, we just feel better about stronger parts.
WB: But my parts meet the requirements at minimum weight. If you'd like stronger, heavier parts, you guys need to change the requirements.
SG: !@#$%^&*(
Hope you guys aren't still mad at me :-)
Humor me with a tangent: I think we should go back to design doing the stress work, for the most part, with some graybeards in stress and their anointed apprentices providing guidance and verification. AFAIK we really don't have the tech tools to do that efficiently, though, and nobody wants to invest the time. Design software needs access to all of the loads, temps, material properties from mil handbook, knockdowns, everything needs to be versioned and traceable, it needs to let you choose from a large catalog of analysis methods with good FEA as a fallback, show you the updated MS and critical load cases in realtime, always move your analysis nodes with the model as you edit it, export 3D models into the stress reports that remain live when the part/loads change so it changes in the next version of the report, etc. You mentioned git elsewhere. It would be so much better to validate stress report updates by viewing a diff. I could go on. Plenty to disrupt in this market, but the incentives aren't there for more than slow incremental change. I think only somebody who's willing to sink the money and open source the result has a chance at creating a revolution in how we design airplanes.
We were mentored by greybeards who worked on the B-47 and 747, and sometimes would still use a slide rule. I received quite a wonderful education from those guys.
I was the first flight controls engineer, that I could tell, who stole time on a PDP-11 to much more efficiently do my work beyond what was practical with a calculator. (I was called on the carpet for that, the manager of the computer center wanted me censured for it, but my boss's boss's boss went to bat for me and told him to fork off.)
It means somebody left out a load or a constraint, or messed up the material properties, which is all another way of saying the math model is wrong.
So when a part doesn't look right but seems to pass the math, be very suspicious of the math! :-)
(wikipedia does not have any other info, so I assumed you went to computer industry pretty much immediately after your aerospace eng degree...)
But we must remember a famous line from a bad movie: "You don't actually think they spend $20,000 on a hammer, $30,000 on a toilet seat, do you?" Given what this aircraft is, what it does, I think it safe to assume a high degree of vagueness and miss-labeling in the non-classified version of the budget.
Use the $5 billion to build housing for homeless people.
We find/print money for things we want to do.
We have a national airborne command post, ultimately, in case the ground based command (the White House) is destroyed or under imminent threat requiring it to be (temporarily or permanently) abandoned. It's used for routine transport, and command and control while being used for routine transport, since it is desirable for it to be near the C-in-C at all times, and it doesn't make much sense not to use it.
Homeless people are outside the circle of social distance and, therefore, social sympathy. Just sayin' the way to lower the cost is to produce more and sell more reprints of the manual at scale.
Every single line of that document would have been double checked multiple times, and literally have 3x more documentation tracking why that piece of top level documentation exists and tracking any changes to it.
The average reading speed is about a page every 1-2 minutes (for non technical texts). 100.000 pages would then be a 800-1600 hour read, which is about half a year for a person spending 8 hours a day just flipping through pages.
Obviously such a manual is probably not meant to be read cover to cover, but even compared to a reference work, it's three times the volume of the entire Encyclopaedia Britannica - a giant amount of information to be easily used.
The information in that manual is probably spread over the domain of 100's of people all performing different roles. I don't think their is any expectation that one single human would read the entirety.
Kinda like your car manual or university textbook. I've never read mine from cover to cover, but it is a useful reference sometimes. Particularly in the case of Air Force One where I doubt an internet search covers most troubleshooting requirements.
Research by thousands of highly trained engineers, compiled by experienced technical writers and illustrators: $83 million.
Research by highly trained engineers: $50M
Printer ink: $32M
Before I worked in aerospace as a structural engineer, I couldn't understand why airplanes are as expensive as they are.
Once I'd worked in aerospace as a structural engineer, I couldn't understand why airplanes aren't more expensive than they are.
I worked in stress analysis, where you'd analyze a part of an airplane to determine whether it fulfilled strength/fatigue life requirements. At this stage, the part's already been designed. So you take the part's material properties, geometry, surface preparation and coatings, machined features, and so on, and check it against anywhere from a few dozen to many thousands of load scenarios. You find the handful of critical load cases, and then you write up the analysis against those. The analysis is intended to be mostly self-contained, readable by any qualified engineer, and serve as a complete treatment of the part's structural sufficiency. It's not much of an exaggeration to say that every single structural component of an airplane is backed up by a stress analysis report running somewhere between 20 and 400 pages, written by one or more degreed engineers making what was, in the U.S. probably an average of $50-60/hr. Airliners are made up of many, many structural parts. The costs of proving the safety of those parts add up quickly.
Now, if I had to guess, instruction manuals are probably a bit cheaper to produce. Tech writers probably aren't as well compensated as engineers. But a lot of engineers' time will be eaten up nonetheless. (I once spent a memorable few weeks (!) proving that a dowel pin on a part that hadn't been machined correctly would not, during installation, shear off, so that existing installation documentation could still be used. Yay. After that I was known as the office dowel pin SME.) Good maintenance documents sweat the details and require input from designers, stress analysts, manufacturing engineers, supply chain representatives, and others. And then as the design evolves -- the design always evolves on the way to delivering a product -- these documents must be tweaked, added to, subtracted from, sometimes entirely rewritten, and those changes must be checked by a variety of folks.
So, $84 million for 100,000 pages might sound like a lot. But for what is in many ways a custom built aircraft, with divergences from the base aircraft that probably run into the millions, to me, it doesn't seem ridiculous.