Tuesday, April 28, 2015

"Drones" are a big category

Uber has announced food delivery in four cities. Uber is building self-driving cars. SpaceX has an autonomous spaceport drone ship. The FAA is (agonizingly slowly) creating rules for autonomous flying drones, while laws for self-driving cars are being handled at the State level. Rolls Royce is building an unmanned cargo ship. FedEx wants to build drone cargo aircraft to air-freight cargo from Asia. Drug smugglers are using unmanned submarines.

What do all these stories have to do with each other? It's that drones are bigger than just the aerial robots that hobbyists and the military use. This is a huge, tectonic change that is going to affect every part of our society that moves physical goods between two points. Amazon is leading in several categories (including inside their own warehouses) but there's so many possible use-cases it should be assumed there will be many niche players. Uber's food delivery service is just the tip of the iceberg.

I think a good mental model for the coming "moving atoms infrastructure" is the Internet. The Internet is a dumb pipe for moving bits over TCP/IP. Once drones because a sufficiently inter-connected network, it will essentially become a dumb pipe for moving packages. The centralized systems currently managed by FedEx and UPS will be replaced by decentralized systems that route packages just like routers treat packets.

This may seem like an unlikely scenario (who's the ARPANET in this scenario?), but I think it will become necessary. As drones take over moving more things, they will start to have to cooperate. UPS might not pick up at your location, so you hire an Uber to take your package to the nearest UPS store. Then there's a hand-off. How does that work? Uber and UPS will have to create a common protocol for exchanging payment, as well as shipping and handling instructions. This protocol gets adopted by others too (maybe Amazon and Wal*Mart use it to accept deliveries at their warehouses from UPS, FedEx, or a new competitor), and pretty soon you've got SMTP for boxes. Even if a company like Amazon wants to be the whole market, they can't. "Moving things" is too big for any one organization, so protocols of cooperation will be necessary.

And the fun part will be when someone creates "Tor for boxes", for ensuring privacy in shipping. There could be even be companies that "ensure discretion" by sending an unmarked drone (aerial or wheeled, depending on your needs) to your location for pick-up and then hand-off the box into the shipping network at a random location to hide it in the stream of commerce. The DEA will have a fun time with that one.

Friday, April 24, 2015

Could Uber make Orbitz obsolete?

Uber is a company that gets people from point A to point B, as quickly, cheaply, and hassle-free as possible. Right now they're doing it with black car and part-time taxi services, and in the future they'll definitely add self-driving fleets of cars to the mix. They may own those self-driving cars, or they may not. I can see a future where Uber licenses their self-driving technology to car manufacturers for minimal cost. Or even for free, as long as Uber's service software is hard-coded into the car as a choice. As long as end-users use the Uber app to call the car, Uber makes money on it eventually.

Right now we think of Uber as a "intra-city transport" company, but as self-driving cars, especially high-speed self-driving cars moving in close formations to take advantage of fuel-efficiency gains, become more common the range of an Uber trip will potentially stretch much further. It's not inconceivable that users will summon a self-driving Uber to take them on a long, overnight road trip so that they're in a distant city by morning. A minivan with the middle and back seats taken out can easily fit a sleeping sofa. (I speak from experience on that point)

At that point, Uber cars will compete with Southwest and Greyhound for intra-continental travel. Short range will always be based on car travel, and coast-to-coast will probably be air-dominant, but there will be a mid-range trip length where it could go either way, especially depending on the particular consumer's sensitivity to price and personal time-value.

At that point, I can see a future where the Uber app asks for a destination and then presents consumers with competing itineraries, all provided by Uber. It can drive you directly from New York to Pittsburgh, or it can drive to you to the airport, arrange for a ticket, and pick you up at the other end in another Uber car. Point to point transport either way, just choose your trade-off between cost and time.

And what role is left for Orbitz or Priceline in this future? Well, hotels and other destinations, I guess. It's not like AirBnB will eat 100% of that market.

Obviously the same logic above applies to Lyft and other Uber competitors, if they think of it. It will be interesting to see which of these companies is first to strike a relationship with an air carrier.

Thursday, April 23, 2015

Google Fi...rmware

Google has launched their mobile service. There's coverage all over the place, but most of it seems to miss the point. Ignore the Wifi, that's mostly irrelevant outside the home. Ignore the pricing. It's a bit cheaper than Verizon in the US, and about the same at T-Mobile, but that isn't the interesting part. The key feature of Google Fi is that it inserts a firmware layer between the end-user and the wireless networks.

The company that controls the end-user relationship, dictates the market. Steve Jobs learned this when he first tried to make a phone and it resulted in the execrable Motorola Rokr. Jobs learned from that experience that he needed to find a mobile carrier desperate enough gain market share that it would let him make the phone the way he wanted, and Cingular was that phone company. Only after the iPhone threatened to eat the entire mobile market did the other carriers turn to Android to stave off irrelevancy.

Well it's happening again, only this time Google has found desperate partners in T-Mobile and Sprint. Where the iPhone was Apple taking control over the user interface and hardware design, Google Fi takes control over the billing relationship and decides which wireless connection to route data over. It moves the control of tech companies deeper into the device, to the layer of the firmware and the SIM card itself, essentially pushing the mobile carrier off of its toehold in the device market. Going forward, the mobile carriers will get revenue when Google Fi decides to route traffic to them, and not otherwise.

With Google Fi, Google controls the wireless service that the user is getting data from, and much becomes possible in the near future. Google Fi will be able to easily integrate future networks, like Google Loon, the SpaceX-Google satellite initiative, or even whitespace or mesh networks. Google will also be able to intelligently route data over multiple networks using Multipath TCP so that Google Fi will (from the end-user's perspective) be faster than any one of the networks that it is made of. Software innovation the carriers never bothered to implement will be allow user-controlled settings to prioritize price, bandwidth, or latency, even routing different services over different networks so that data backup uses whatever is cheapest while gaming uses whatever has the lowest ping.

And that software/firmware-enabled flexibility is the key. Carriers today thrive on locking up users for years at a time. On Google Fi the carriers will have to compete every second of every day against the auction process embedded in Google Fi's firmware layer. And not just compete against each other, but also other forms of transit such as the Wifi, Facebook's drones, and the balloon and satellite technology mentioned above. Every ounce of margin is going to be sucked out of their business, eventually resulting in lower prices for consumers. And of course Google's partners Sprint and T-Mobile have to know this, but at this point they've got nothing to lose.

Eventually the carriers as they exist today will die away as consumer-facing brands. They might hold on to enterprise service business, or they might not. Hard to say. The network of towers they've build though will slowly become the dumb pipes that they are, and subject to unrelenting moment-to-moment competition. Depending on how Google Loon, SpaceX, or Facebook's drones interact with Bitcoin-powered mesh networks I'm not even sure they'll survive as firms. The "bright side" scenario for them is that mesh networks don't succeed and they become business-facing network managers who only earn whatever profits their spectrum ownership allows for. The "not as good" scenario is that mesh-networks do succeed, and they transition to an existence similar to Linksys or ASUS, just selling rebadged OEM hardware to end-users.

Expect Apple and Microsoft to eventually make similar services for iOS and Windows. Apple loves controlling the consumer experience, and this will only make it easier. Microsoft will eventually follow Google and Apple to maintain feature parity, as it usually does in mobile. At that point the mobile-OS companies will have taken full control of their customers, the orifices will be declawed, and consumers should win in the sense that they'll get more bandwidth for less money.

UPDATE: Byrne Hobart sent me the link to this Joel on Software Strategy Letter V. A key quote from the letter is:

Smart companies try to commoditize their products' complements.

Indeed. Google's key product is Search (Advertising), and its complement is bandwidth. The profitable mobile carriers (Verizon and AT&T) are resisting being commoditized, but Sprint and T-Mobile are willing to risk it. Google Fi, by making bandwidth "just work" for the consumer commoditizes the connection. Google Loon and the other initiatives mentioned above will continue that trend.

Wednesday, April 22, 2015

ULA's Integrated Vehicle Fluids

Frank Zegler of ULA provides some background here, and with a follow-up here, on the development of their Integrated Vehicle Fluids (aka, internal combustion engine (ICE)) for the Centaur upper stage. A question and answer at Stack Exchange explains the practical benefit.

This is really great stuff, and may be what keeps ULA "in the space business game" once the expendable Atlas rockets are retired. No one else has technology like this that I'm aware of, and ULA's in-space "rocket trucks" could become the deuce-and-a-half of the coming space age. The increased mission duration and number of burns is particularly relevant for anything beyond LEO - such as missions to the Moon or an asteroid.

The shape of the business in the 2020-2030 range seems to be: SpaceX launch, ULA cargo movers throughout Cis-Lunar space, and Bigelow-leased campers and destinations (passenger cars, space stations, and lunar/asteroid habitats). Even companies developing their own in-space technology (such as Planetary Resources) might decide to not duplicate R&D efforts and rely on ULA for moving things around.

Also, take a moment to appreciate the irony of the fact that Elon Musk (by making ULA's Atlas rockets uncompetitive in the launch market) has forced them to create a new market for internal combustion engines.

Into the Black

A post from Centauri Dreams explores what a nomadic colony in the Oort Cloud might look like.

The tl;dr to his idea is: small bands (~25 individuals) tending to really big mirrors collecting starlight for energy, traveling in groups that total ~500 individuals spread out over an area the size of the continental United States. (It has to be this large because starlight in the void between solar systems is so diffuse that the solar energy collectors are enormous)

Personally, I think this sounds like a lousy idea.

Socially, small bands of 25 people may get along, or they might all kill each other. Humans need a bit more space than that. Further, such small groups would require that everyone be an extreme generalist so that any one person's death doesn't result in the death of the colony (you never want to hear "Only Bob knew how to fix the air regenerator!" 10,000 AU from Earth's atmosphere). Modern human society is only possible with specialization, so I'm not even sure this level of generalization is feasible.

(Of course these future space colonists would have access to very sophisticated and intelligent software that could probably talk them through repairs, but that presents a deeper risk if no one on the colony really understands personally how the systems work or interact)

Also, to protect the colony from interstellar radiation, you're going to need a very thick water-ice jacket. There's plenty of water in the Oort Cloud, but internal volume, mass, and surface area all have fixed ratios. If you're trying to accelerate a volume big enough for 500 people out of the solar system, a single volume with a single water-ice shell will have much lower mass than many smaller volumes each having their own shell.

From a colony safety point of view of course there's a risk to having everyone in a single volume, but the tradeoff is you have more people and resources in one place to respond to emergencies too. And bigger systems have bigger buffers in almost every parameter, ceteris paribus.

Of course, what's the point of being out in the Oort Cloud in the first place? It's so far removed from planetary masses and solar energy that no civilization will ever be as comfortable there as it would be inside Jupiter's orbit. The main asteroid belt make sense as a destination for permanent settlement - the Oort Cloud, less so.

As I see it, there's only two reasons to be in the Oort Cloud: to collect resources for use by the solar civilization, or as a stepping stone towards the next solar system. And neither of these cases require long term, fully sustainable colonies. The first one requires the deep space equivalent of an offshore oil and gas platform, and the latter the deep space version of a cruise ship. Both uses have predictable mission lengths, and thus huge solar collectors are unnecessary. Just bring a nuclear battery.

What might a colony ship look like? Well, if resources are scarce or expensive in the solar system, a group of colonists could put together a mission that only has enough fuel and material to reach the Oort Cloud, and then scavenge an Oort Cloud object to provide the rest of what they need. A single ship, surrounded in a thick water-ice shell, with a nuclear core keeping everything warm. Rotate the thing to add gravity. A few hundred colonists could then use the deuterium from the comet to accelerate their little seed pod towards the next star, where they could then settle or resupply and keep going. The nuclear battery would run out of fissionable elements eventually of course, but you just plan for that and make sure you have twice as much as you think you need to make it to Alpha Centauri.

Sound risky? Sure, but then so were the wagon trains to California. Or the rafts used to colonize Polynesia. People have taken greater risks in the past to find a new home, and they will again in the future. And at least these colonists can pack tens of thousands DNA samples with them to avoid inbreeding en route or at their destination.

Tuesday, April 14, 2015

The Vulcan Rocket

With SpaceX's Falcon family of rockets delivering on cheap, reliable access to orbit, and with the real possibility of even cheaper (reusable) rockets in the near future, the Delta and Atlas rockets operated by the United Launch Alliance (ULA) are obsolete. There's no future for a launch service that cannot compete at the new price levels.

Unfortunately for ULA, once the Falcon rockets are reusable, there's no way an expendable rocket can compete on price. ULA must find a way to make their rockets reusable. But landing rockets (in a way that they are immediately available for a new launch, as opposed to needing months of expensive refurbishment like the Space Shuttle or its SRBs) is a very hard technical problem, and ULA is way, way behind SpaceX in solving it.

Which brings us to today's news: ULA has announced the Vulcan rocket, which they hope will fly within four years.

Cheap, reusable rockets don't just come together within four years though, and ULA has been sitting on their hands for years hoping SpaceX would just go away. In order to meet this timeline, ULA has basically given up on making an engine or learning how to land a rocket.

The engines for the Vulcan are going to be made by Blue Origin, Jeff Bezos' rocket company that has been working on building a reusable system for at least ten years now. This is consistent at least with the current Atlas rocket - ULA doesn't make those engines either, but buys them from the Russians. Despite the storied history of Boeing and Lockheed, their institutional expertise at building rocket engines has been allowed to atrophy and die over the last several decades, and Vulcan will not revive it.

Secondly, the Vulcan won't even land, Instead its engine module (the most expensive part) will detach from the main body, use an inflatable aerobrake shield to reenter the atmosphere, then parachutes, and then be snatched out of the air by a really big helicopter before it crashes into the ground. If that sounds more like a circus trick than a reliable mode of transport, you're not wrong. The comparison between this system, and the Falcon 9's powered and steerable descents, could not be more amusing. ULA is desperate to save and reuse the main boost engines, and the best they can do is this Rube Goldberg contraption.

And despite all that effort, ULA's cost target for a Vulcan launch is $100 million. And that's four years from now, assuming everything is on time and on budget. Compare that to $61 million for a Falcon 9 today or $90 million for the (upcoming but even more capable) Falcon Heavy. And those costs will be cut in half, at least, once SpaceX licks the reusability problem.

This desperate bid for technical relevance by ULA would be hilarious if it wasn't such an inglorious end to America's first space-age companies.

Thursday, April 9, 2015

The EM Drive

Since following technology is a hobby for me, I follow the technology development that's fun to think about succeeding, even if the odds are low.

One of my favorites is the EM Drive. It's a proposed thruster for spacecraft which claims to produce thrust without reaction mass. It turns pure electricity into linear thrust, without having to throw anything out the back of the ship. There are similar research projects around the world looking into what seems to be the same phenomena, including Paul March's work on a Mach-Lorentz thruster, the Cannae drive invented by Guido Fetta, and replication work of the EM Drive being done by some researchers in China.

If the EM Drive or something like it ever works, it would initially not help us reach orbit. The thrust is too low to escape Earth's gravity. You'd still need chemical rockets to reach space. Once you're in space though, things change. A small but steady thrust can add up to very high velocity, helping making trips to Mars or further out much faster. Instead of reaching the outer planets in years, it would only take a few months, much like a ocean voyage during the age of tall ships. Further, since you don't need to bring reaction mass with you (and most of a space craft's mass is currently devoted reaction mass and/or oxidizer) it opens up the mass budget of any space mission to add more crew or science payloads.

Here's a current summary of where EM Drive research is sitting. (Be sure to reach the comments. Next Big Future is the rare website where the comments are almost always value-ad) The very short summary is that very small thrusts have been measured at NASA facilities, and now they're trying to scale it up a bit. If they can get it to scale up even a little it will be immediately useful on satellites and space stations for help maintaining their orbit. And then with a little more scaling, the whole solar system becomes available for exploration and settlement.

(And if it can continue to scale, further out you might eventually get that X-34 Landspeeder).

Wednesday, April 8, 2015

Private Spaceships

Forbes has a gallery of the private spacecraft currently in development.

Including the Worldview balloon and Bigelow's projects are category errors (balloons can never reach space, and without propulsion Bigelow modules aren't "craft"), but they're still cool so I'll allow it.

The good news is that none of these projects are paper rockets. They're all real hardware with real teams. That's no guarantee they will all succeed, but this commercial activity is a refreshing change from the endless SLS and Orion development being done by NASA recently.

Wednesday, April 1, 2015

Lockheed's bid for relevance

Lockheed and Boeing are the grand dames of American aerospace, having long gobbled up their competitors. Both firms have grown comfortable in the lack of competition, and have even eliminated each other as competitors in the space launch business by forming the United Launch Alliance partnership to operate their Delta and Atlas rockets. Their business model is one of government cost-plus contracts with limited competition and a sprawling collection of sub-contractors.

Into this landscape came SpaceX a few years ago, and they have eschewed cost-plus contracts for fixed price offerings with the price listed on their website (a first in space launch), sprawling webs of sub-contractors for vertical integration, and stable technology states for constantly improving systems. Together these innovations have to date reduced the cost of reaching orbit by nearly 2/3rds, and near-term innovations like the Falcon Heavy and reusable rockets threaten to lower costs even further. These cost reductions have allowed SpaceX to capture significant portions of the private satellite launch business and also large contracts from NASA.

I suspect that when SpaceX first came onto the scene the executives at Lockheed and Boeing were skeptical they would succeed and probably figured they could just wait for SpaceX to go bankrupt and then go back to business any normal. Any disruption to their business would likely be temporary.

However likely SpaceX's failure was ten years ago though, it is now very unlikely to fail in the near term. The Falcon 9 is a proven system, and SpaceX doesn't even need the Falcon Heavy or reusable rockets to continue to dominate the launch market. And since the Falcon Heavy is just three of the Falcon 9's proven launch cores strapped together, odds of its success are good. Even the heavy lift launch market is under threat from SpaceX.

So what should Lockheed and Boeing do? The fact is that their rockets cannot compete with the Falcon rockets as the Falcon rockets exist today. If the Falcon rockets become reusable and lower prices another order of magnitude, it's well and truly game over.

This background understanding is necessary to evaluate the announcement from Lockheed about their Jupiter spacecraft and Exoliner cargo vessel. This project, and its announcement, is Lockheed conceding that SpaceX has succeeded in changing space business and that there's no going back. Consider the following:

  • Lockheed is developing Jupiter and Exoliner with its own capital. They didn't wait for a NASA or DOD contract to pay them to do it.
  • The Jupiter spacecraft is designed for deep space missions, which competes directly with the Orion spacecraft under development with NASA.
  • Lockheed is talking about going to the Moon, which is not on NASA's agenda at all. This marketing is obviously directed at foreign governments and private initiatives. 
  • Lockheed is marketing the (relatively) low cost of its system. SpaceX's choice to advertise and sell on price has infected Lockheed. 
  • Jupiter is refuelable in orbit. This is entirely new technology that no one (not even SpaceX) has announced support for.
Each of these changes individually is huge. Together they reflect the tectonic shifts going on in the space business these days. After 70 years of government programs, space is becoming a "normal" technology business, with competing, innovative firms lowering costs and improving quality while selling their wares to all comers.

The abilities of the Jupiter craft in particular are fascinating and suggest where Lockheed sees the future of the space business. Lockheed has designed Jupiter as a generic "train engine" for Earth orbit, Cislunar space, and beyond. They no doubt are seeing the Bigelow modules that are coming and have thought to themselves "Those things will need help getting around in orbit, or to the Moon". They've seen the future of multiple space stations, asteroid mining, Lunar ice mining, and who knows whatever other things cheap Falcon rockets will put into orbit, and asked themselves "We can't compete on launch any time soon, so where can we compete?" Their answer, at the moment, is moving things around in space once the launch part is over with. It's a great answer, and will improve the odds that Lockheed plays a meaningful part in our solar future.

Now for a bit of speculation on future announcements. The Jupiter craft is refuelable in orbit. At the moment Lockheed is just talking about it being refueled whenever it docks with a launching rocket to pick up new cargo. This will work, but it's not the best method. Lockheed has to be thinking about building fuel stations in orbit. These fuel stations could then be refilled by any third party on a lowest-cost bid. Even SpaceX could get into the game of launching fuel to the on-orbit stations, and eventually the asteroid and Lunar prospecting companies could compete for this business. Lockheed then wouldn't just own all the trucks moving cargo around in orbit, but the gas stations too. It would be a good business to be in, as satellites, space stations, and space craft of all kinds will (barring some breakthrough in physics) always need fuel to operate in space.