Hacker News Reader

Comment by alexasmyths | Hacker News Reader

+10.

This should be done in many more places.

Though with 'cost plus' contracts... providers are incented to create crazy new things for every problem.

If there weren't moral questions around this - it would be amazing to see a NATO forces 'hackathon' where people create top gear/tech for soldiers with off-the-shelf components and software.

I have a funny feeling that 'if they put their hearts to it' - there'd be no competition in terms of what could be created.

Imagine the SpaceX of 'jet fighters' - 2x the capability, 5% the cost.

Anyhow, obvious moral issues there ... just a thought.

View on HN

"Hacking for Defense" is a step in this direction E.g http://hacking4defense.stanford.edu

"Defense" :) always have to laugh about that part

We have a "Department of Defense" as well as one for "Homeland Security". Cognitive dissonance is a funny thing.

Well, you have the issue of EMP hardening. Not sure if that is a problem in submarines. But elsewhere, you have a serious defensive flaw if an EMP attack can disable military hardware.

Is there an EMP capable torpedo though? They seem useless in water. Other than the nuclear ones of course...

I think the durability of an Xbox controller is of no consequence against a nuclear torpedo.

I doubt an EMP torpedo would exist targeting submarines underwater. Water is just crazy good at dampening electromagnetic waves. And if you can score a direct hit, the EMP device's weight in extra explosives is cheaper and more destructive.

A submarine's hull acts as a faraday cage. It's inherently quite well shielded against EMPs.

How many COTS parts does a SpaceX rocket use?

I guess there's no way to know for outsiders, but I've heard SpaceX buys a lot from McMaster-Carr. Parts for rocket itself or other tools used to build/test rockets? I have no idea.

I've been reading the bio of Musk and it discusses that they progressively replace COTS as much as possible, building and designing in house because in the end, THAT is cheaper.

The Musk biography is a pretty good source for that question. From what I remember, SpaceX produces a rather extreme fraction of its parts in-house (something like 80%–it's been a while). What they don't create themselves, they often source from the consumer market. They avoid the tradition space industry supplies wherever possible because they have had success with off-the-shelve parts available for a fraction of the costs.

I believe the book gave a radio module as an example. It was available for $120,000 at "space-rated" quality, but they bought something at RadioShack for around $3,000, and later replaced it with an in-house part that was smaller/lighter/more power efficient.

To be fair 'space rated' does have meaning.

There is 'off the shelf', 'automotive', 'aeronautic', 'unmanned space' and 'manned space' quality gear.

If you think about it - when you're flying a plane, the 'tolerances' for everything just has to be higher - there's much more at risk.

In a complex system, there are a lot of single points of failure that can cause haywire.

Unsurprisingly, the tolerances for 'unmanned' vs. 'manned' space flight are quite different - and for good reason.

Perhaps the economics are not always quite right, it may make sense to make some small probe out of 'cheap gear' - but there is of course added risk, and usually, you 'only get one launch' attempt.

Finally - there's a reason a 'space grade' radio might cost $13K, and that's because they don't make very many :).

I interned on the Space Shuttle robot arm project - we made these aluminum 'handles' for the arm to grab onto. I mean - really, really basic stuff. $200K each. Engineering, overhead, testing, sales yada yada - we only ever made a few dozen.

>we made these aluminum 'handles' for the arm to grab onto. I mean - really, really basic stuff. $200K each. Engineering, overhead, testing, sales yada yada - we only ever made a few dozen.

That sounds ridiculous. Did your coworkers have a strong opinion of that, or not care?

It's not ridiculous at all, it just seems that way.

It takes a lot of work and investment to design a 'space grade product'.

When you only ever sell 50 of them - total, well the price will be high.

Imagine if only 500 iPhones were ever sold - how much would they cost in order for Apple to make a profit?

It's just the weird economics of small batch products that makes the price seem bonkers, there's no volume to spread the fixed costs over.

IPhone has a lot of R&D required for it to do what it does.

How much R&D is reasonable for some "really simple handles?"

Requirements analysis, systems analysis and research (when you do things in space absolutely everything needs to be looked at), materials design and test (materials behave differently in space). Fixtures? How are they affixed to satelites? You can't just 'glue' them. The type of 'screws' are pretty important. This might require a lot of back and forth.

Each step of the process involves internal review and review by the customer - i.e. Japanese space agency, and can take weeks.

Testing requires a massive 'clean room' which is expensive to maintain. And you don't just make these things anywhere - tolerances have to be perfect.

Not only is it tough engineering, it's a lot of bureaucracy.

And then overhead: IT, support, software, flights, hotels yada yada yada. It adds up.

You've said why it is expensive. Not why it needs to be expensive.

That's literally a salary of 2 upper-grade engineers for a year (for a piece).

There's literally _no way_ it had been done in the most efficient manner.

Though there is a lot of bureaucracy, you'd be surprised at how much work it is.

Consider how much a massive 'clean room' costs to operate.

Consider how much gear it takes to simulate the operation of massive robot arm in a 0G environment, when we live in a 1G environment.

Consider what happens when a 1 ton arm, attached to a 20 ton structure, applies toque to a 1/2 ton satellite... when the satellite is moving ...

Consider how much it costs to send 4 Engineers, a PM and a Bus Dev to Japan for three weeks for talks with Japanese Space Agency? Several times?

Consider the lawyers bill - Japanese, Canadian and American ones. Translators. Insurance estimates. Liabilities.

FYI - did you know that every single thing that goes into space is tracked? Every single screw had an ID number. It's original manufacturer, shipping, the date it was put on the station, by whom, when it was removed - etc etc. Every goddam screw :).

Now think of how much software and admin is needed to track all that.

And now consider that if one single manager, at one single subcontractor opens an 'issue report' anywhere in the world - that the 'countdown' for a shuttle launch was postponed? (Happened to my boss, opened a tiny change request on Admin Software in Toronto - went for lunch - Shuttle countdown froze, panic and hilarity ensue). And the overhead costs associated with that.

Space is expensive, especially 'manned space'.

Vert complex, one-of-a-kind system - where nothing is allowed to go wrong.

And we've managed to end up with shuttle with all of that bureaucracy.

I understand the complexity of the issue at hand. I can.

Nevertheless simple handles going for 200K a piece is a vast budget. Any proper business would quickly open up an internal department to "do the handles" since it both allows them to grow for free (you get additional specialists and the handles) and allows to use the same engineers for something else. More importantly testing becomes less of an issue since now you may easily do integrated testing of several disjoined components at once.

So no - I understand where the expenses come from. But it doesn't mean that structure is in _any_ way motivated.

The idea of subcontracting and every little piece of hardware being done by a separate little entity doesn't seem to work both in terms of pricing and in terms of the amount of work it would be doing be it a merged entity.

Case in point, SpaceX has demonstrated that applying standard business optimization tactics to space hardware manufacturing works.

Mostly. Note that EELV and NASA launches by SpaceX have an extra paperwork charge, which is as big as $20mm. One hopes that eventually the US government will be able to take advantage of the same, cheaper service that commercial launches buy.

"easily do integrated testing of several disjoined components at once."

It just doesn't work that way.

It's an area of hyper-specialization.

Everything made for space is a 'one off', there really are no economies of focus or scale. Maybe in launch.

There just aren't 'lines of business' in space really.

Once you walk in the situation and spend a few days there, and see what's going on, you get it.

I haven't even scratched the surface of it.

Every process is recorded, needs to be archived.

Every single piece of software hardware ever used needs to be 'kept around' in case something goes wrong. 20 year old computers, floppy disks, have to be managed and 'kept up' in case something from way back needs to be tested.

Purchasing requirements/limitations. Political control (governments pay for this, so they set rules, change them). Security clearances. Accounting standards. Process audits. Insurance audits. Hyper-specialization HR impossibilities (like 10 guys in the world qualified to do ABC activity).

You must wonder why it costs '100M dollars' to film a big movie, have a look at the detailed budgets. A lot comes out of the woodwork.

I think I'm really starting to understand more about all the reasons why SpaceX can get stuff in space for a fraction of the cost of Nasa in less than a decade in business

NASA doesn’t build rockets. You mean Lockheed/Boeing/ATK.

Basically yes. The NASA space shuttle was conceived by NASA, chosen by Nixon among a group of possible projects, and then the tank built by Lockheed, orbiter by ATK, etc. If anything, many of the above problems wouldn't just go away, some of them would become even more costly.

Maybe to avoid issues like this. https://www.wired.com/2010/11/1110mars-climate-observer-repo...

Not as bad a mistake as the Challenger Disaster (engineers knew it would happen, but management ignored them).

Perhaps you should hawk your minimally-tested, 'just trust us', generic pre-designed handles made out of cheap steel to NASA, and tell them what a bundle they'd save. If it works then you'll be on a gravy train.

Well, a limited-run gravy train, for 50 or so units.

I appreciate some good snark, but that's more or less exactly what SpaceX is doing, although their handles are slightly more complicated.

By using off-the-shelf hardware, and reusing the first stage, they're on a good trajectory to undercut previews launch system by almost an order of magnitude. And NASA trusts them enough to outsource even manned flight to them.

Don't you have (shouldn't you have) a "certified parts library" so you don't need to go through the process of studying which material is needed for a handle, use standard and pre-certified screws to affix them, which you already know won't come off due to vibrations during launch, and won't accidentally fuse once in space (https://www.quora.com/If-two-metals-touch-in-space-they-fuse...). Basically do it right, so you only do it once, like software libraries (in an ideal world). Unless you really really have a damn good reason to not pick an existing handle from the library.

What makes you think people would NOT have consulted paat experience? Given this is space parts, there simply may not have been enough experience.

I can easily believe that.

Next time you're sat in a meeting with 5 other highly paid software developers, discussing a two line bugfix, think about how much that bugfix will have cost.

I would love to see an Outlook plugin that estimates meeting costs using salary band information for attendees. It would help keep rambling to a minimum.

See: https://en.wikipedia.org/wiki/Product_binning for something similar related to computers.

A lot of R

Reply on news.ycombinator.com