AvWeek reporting 9(!) new SpaceX contracts, 2-3 Heavies

[macpacheco]

What, is Ariane now some sort of "luxury launcher"? Does it have leather-upholstered farings and shiny chrome SRBs?

Before F9 (v.1.1), the equation was either cheap, risky launches on Proton/Zenit or expensive, reliable launches on Ariane/Atlas. Now Falcon appears to be both cheap and reliable, which is why they are making money hand over fist. Unfortunately for competition, none of SpaceX's competitors are able to do both cheap and reliable (though Long March is catching up).

No Ariane like ULA is a pork barrel based business. As such it employs probably at least twice as many people throughout its supply chain than a similarly sized SpaceX type supply chain, plus one of ULA problems is its dependence on many subcontractors that have a de facto monopoly over what they supply ULA with, giving them unfair pricing power, I wouldn't be surprised if Ariane has the same problem. Lots of limitations due to labor union deals, most pork barrel programs (ULA/Ariane included) are JOBS programs. So eliminating a job for cost cutting = bad politics with those that give you money (US Congress / European counterparts).
But where you read Luxury, you should read ultra conservative, unwilling to take chances with anything out of the box thinking, due to internal resistance (specially internal/supplier intertia). The SpaceX model leads to a much lower cost of effect a change in its products vs ULA/Ariane.
It takes a culture that is constantly innovating to keep innovating. Once you get afraid of change, its really hard to break the bonds that slow down innovation.

[Jim]

No Ariane like ULA is a pork barrel based business. As such it employs probably at least twice as many people throughout its supply chain than a similarly sized SpaceX type supply chain, plus one of ULA problems is its dependence on many subcontractors that have a de facto monopoly over what they supply ULA with, giving them unfair pricing power, I wouldn't be surprised if Ariane has the same problem. Lots of limitations due to labor union deals, most pork barrel programs (ULA/Ariane included) are JOBS programs. So eliminating a job for cost cutting = bad politics with those that give you money (US Congress / European counterparts).
But where you read Luxury, you should read ultra conservative, unwilling to take chances with anything out of the box thinking, due to internal resistance (specially internal/supplier intertia). The SpaceX model leads to a much lower cost of effect a change in its products vs ULA/Ariane.
It takes a culture that is constantly innovating to keep innovating. Once you get afraid of change, its really hard to break the bonds that slow down innovation.

I believe this is incorrect and unsupported and you may be repeating yourself.

[ChrisWilson68]

But not too small to make F9 look pretty darn good. 100% successful >primary< payload delivery on the first twelve launches is impressive given that you would expect the highest failure rate early in the life cycle with teething problems. Neither Pegasus nor Taurus achieved that record, and they were less complex designs. No wonder customers are lining up.

Neither did Ariane 5, either! Several early failures.

FWIW, a good way of guesstimating the reliability in the case of no main payload failures (because nothing is 100% reliable) is to assume half a failure... So 12 flights is 12/12.5... About 96% reliability, conservatively speaking. Just a guesstimate, though. Could also take a Bayesian approach which would give much the same answer.

12 missions without a failure is an 87.4% reliability at 80% confidence. It's a 94.4% reliability at 50% confidence level.

Sigh.  Does nobody here understand statistics beyond a basic undergraduate course?

Any estimate of reliability has to have an assumption about prior probabilities.  You can't just say X events with Y outcome means Z probability with Q confidence.

People so often just grab a formula and plug in numbers without finding out the assumptions in that formula.  What prior probability distribution is assumed by the formula you plugged numbers into to get this answer?

I get really tired of seeing statistics misused because people don't understand that there are assumptions behind the formulas and it's necessary to examine those assumptions before coming to a conclusion.

In this case, we have industry-wide experience that we could use to give a more accurate prior probability distribution than some simplistic distribution that is probably built into the formula that these numbers come from.

Here's an example to illustrate the problem.  Suppose I toss a coin 10 times and it comes up heads 10 times.  What are the odds the coin is a two-headed coin?

The answer is it depends on the assumption about how many two-headed coins there are in the bag from which I selected the coins.  If the bag is half two-headed coins, it's very likely the tested coin is two-headed.  But if there are a million coins in the bag and only one is two-headed, it's very likely that the coin is fair and it was just a fluke that I happened to get 10 heads.

The number of two-headed coins in the bag gives us the prior probability that the selected coin is two-headed, before we've done any testing.  Each test we do (a flip of the coin) allows us to refine the probability, but it's all based on the prior probability.

If we know nothing about the bag and how many two-headed coins are in it, we might just assume half are two-headed and make a calculation based on that.  But if we've examined several other coins in the bag, we have some better data to go on in choosing a prior probability.

In the case of rockets its even more complicated because unlike a coin, there is the issue of a distribution of failure rates.  For example, it may be 10% likely you have a failure that shows up with 100% likelihood on each launch (a design flaw), 10% likely you have a failure that shows up with 50% likelihood, and 10% likely you have a failure that shows up with 2% likelihood.  That distribution of kinds of failures and how long they typically take to show up affects the confidence calculation for launches going forward on a new rocket.

Oversimplifying things by using a formula with simplified prior probability distribution assumptions just gives unjustified confidence in the results it offers.

[Chris Bergin]

Back on topic people.

[Nilof]

One of the key issues to think about is that of fixed vs marginal costs. Afaik, there is no organization that sells launches below marginal costs. Government subventions have more to do with making sure their launchers break even in the case where they don't launch as often as they would like.

Material and propellent costs to build rockets if vertically integrated are a very small part of the total costs. The cost of launching them overwhelmingly consists of wages. Since qualified labor generally is not something you can hire on an on/off basis, these wages are fixed costs. Launching twice as many rockets with the same infrastructure and labor force would only cost marginally more to the organization manufacturing and launching the rockets, even if the system is completely expendable.

The key issue is keeping track of the costs involved in maintaining a production line running, and gauging how big your labor force has to be. That and streamlining your subcontractor supply line.

[oldAtlas_Eguy]

No matter how many launches you do off a single pad there will be some of your workfoce idle due to inefficiencies in processing flows that use specalties at only some points in the flow.  Parralelling flows makes more usage but unless the flows number more than two there will still be some labor waste.  having 2 pads in a locatiion and two parrallel time offest flows such that they are like one week apart will almost certainly make maximum use of labor pool. This will be the case once LC39A is operational. 

So once this occurs and reusability is thrown in I wonder just how much of a price drop will occur.  It is possible that SpaceX's prices for these new contracts may be for less prices than the posted amounts because they are for NET 2016 and SpaceX will have 2 Cape pads with possibly Boca Chica and VAFB in operation.