When less is more

Sometimes less is more. This also applies to data. Too long ago to say when without sounding ancient, I came upon a holy grail of launch price data. Many Bothans died to bring us this information. The more extensive, second spreadsheet, not the first one. We had known about the existence of the data, shown a sample on screen only to be told it was “look, but do not touch.” The data was used to shore up the credibility of other analyses, meaning charging more, explaining the difficulty in sharing. There were Delta’s and every kind of Atlas ever, and enough one-off launches here and there to make it look like launch customers had options galore.

Of course, it was mostly noise, not signal. At the time, I did not appreciate this immediately, contentedly twisting and turning the data. Then realizing that was five days of my life that I will never get back. This was not the first time this would happen.

Looking for useful patterns in all the parts that broke on a Shuttle orbiter between flights repeated the experience. Except now it was months. Databases had just gone online in a primitive format (think flip-phone vs. smartphone), but there was a lot of microfiche involved too. The whir, the hum, the cartridge loading and spinning. It was the comforting sounds of data being dug up. One shovel at a time. Everyone should do this once in their lives, so they can relate when they see that movie where the new owners dig up the old news at the town library about what really happened in their house back in ’52. Curiously, my mentor predicted the futility of the effort based on a simple premise – if there was a pattern, someone else would have found it by now.

We are explorers, right? So there must be something buried in all those numbers, and we will find it. There must be a pattern. Well, not always. And if you are big on patterns, there is even a bias for that (pareidolia).

The latter, that abundance of old data, is an excellent way to predict the past.

I distribute a short document called “The State of Play” mostly graphs, built from public data, always searching for useful patterns. Cost data tends to be rich territory if you are familiar with where the bodies are buried and how they are buried. Other data can quickly devolve into noise, which may be a good thing. Scattered data could indicate a changing landscape, which is welcome. In business, this would be the storm before the market shakeout. A lot of innovation, new technology and learning occur before the shakeout.

With this in mind, I am simplifying what used to be six charts about launch pricing down to two. Did I really need to keep a Dnepr rocket on that chart? Or every company that says they will launch in 2021, even when not showing any hardware (yet)? Especially so, did I really need all those old data points? The latter, that abundance of old data, is an excellent way to predict the past.

First, to look ahead, lets just look at recent launch awards. SpaceX is all over the place, from small the heavy launch offerings, with relative immunity to variation by scale. A reusable Falcon 9 at below internet pricing (the IXPE mission) even beat out launching that tiny payload on a Pegasus. The Electron would not have sufficed, and LauncherOne was not available at the time. But it’s not quite over on the small end, as Astra, were it to come online at the advertised pricing, will give pause to anyone thinking SpaceX rideshare (or Electron, or LauncherOne). The Firefly Alpha and the Relativity Space Terran 1 are also in test and hardware heavy.

Oddly, with the data out there for all to see, ULA and Arianespace are shooting well high of the apparent target, SpaceX. Quite simply, a sight (not slight) adjustment will be required to compete commercially with SpaceX and to close any business case. And that’s not counting a toxic, inhospitable Proton, losing business for years, but not to be discounted entirely just yet.

It could be a while again before we see that pretty straight line fitted to data points not too far above or below. I’m removing the line for now too. Less is more, and the kind of disruption that causes loss of signal may be the best thing to happen to launch in a very long time.

P.S. My earlier entry on this topic is “Here there be dragons.” I will also add these graphs to the main menu, keeping them updated vs. the version any date here.

Also see:

• ULA’s Rocketbuilder does not provide pricing for some years now but is still online.
• The NASA Launch Services Program Performance Website comes in handy as a cross check with other data.

Update 11/30/2021: Replaced the graphs after fixing an error in one of the data points (Astra).