A thread of 20 X-band synthetic aperture radar designs I have known…

1. Umbra 🇺🇸 2. Capella 🇺🇸

Recently we announced 16cm resolution data, the highest resolution SAR ever released. Today, three examples of that data have been released freely and openly:

Let me explain what the *real* story is here, which I don't think is obvious... https://t.co/XmixOKODQ5…og.s3-website.us-west-2.amazonaws.com/?prefix=sar-da…
Most people (naturally) seize on the fact that we can produce higher resolution data than anyone else, because they equate resolution with quality and utility in their mind. And tbf it is pretty cool! But what it demonstrates is more nuanced than just "highest resolution data."

The reason satellite imagery companies can’t *also* do analytics well isn’t because they lack the technical talent, it’s because they lack the specialized sales, marketing, and support to build *products* not just algorithms. It’s not impossible, it’s just not a good bet. If 10 startups rush into a new niche opened up by cheaper/faster/better satellite imagery, maybe 3 survive and 1 dominates. You’re basically betting that your part time gig will be that 1 while your competitors are full time on it.

1/?

Maybe I’m nuts but I truly believe that every satellite imagery provider has an ethical and fiduciary duty to give away their archive for free.

…allow me to explain: Ethical: what single act would have a greater impact on climate mitigation/adaptation work than the simultaneous release of all commercial archival satellite imagery? Decades of global environmental change and human activity suddenly unlocked. A new baseline.

Hey everyone! I wrote a quick essay on why so many of the new products I see getting launched in the satellite imagery industry are off the mark. Would love to learn about counter-examples or have a respectful debate with anyone who disagrees. joemorrison.substack.com/p/nobody-wants… Basic summary: people think you can build an algorithm that does a simple thing really well, use it to generate a data feed, and sell that data feed as a subscription. It sounds really appealing. One problem: it doesn’t work. Like, ever. (There are exceptions, but not many).

Ain’t no such thing as a generic change detection algorithm for satellite imagery. It has been tried many times. Never seen it work.

One reason is that every use case defines “change” a little differently. One man’s change is a another man’s noise. Some dumb examples: 1. If you are a city planner looking to quantify urban forest canopy change, you specifically don’t want false positives caused by seasonal changes (lots of trees lose their leaves in winter, doesn’t mean you’ve lost canopy). Changes in grass/shrub/garden/etc. are also noise.

Let’s talk a little bit about what the future of the satellite imagery industry looks like.

Can you code? No? Well, then, it doesn’t look like you. There’s a tectonic shift occurring that is shifting the core business case for satellite imagery providers from mapping (large, exquisite mosaics) to monitoring (deep persistent time series).

Along with that shift, there’s a changing set of customers with new behaviors.

1/4 I feel strongly that there are only two salient business models for satellite imagery providers. Choose one, nail it, and you can be successful.

Try to do both well at the same time, and you’ll get ripped in half. And I’ll probably tweet mean things about you. 2/4 The first model is vertical integration. Proprietary products and services built on top of proprietary data. Extremely defensible, but risky because of concentration in a few markets. Goal: monopolize distribution to one or a few key markets.

Sometimes I feel like I’m going crazy as I watch every single other satellite imagery company investing in their own analytics team. Am I the fool?

Quick thread on why I still think they’re making a mistake and how I might be wrong 👇 First, my assumptions:
1. There is artificially suppressed commercial demand for satellite imagery because of high costs, restrictive licenses, limited capacity, and legacy sales cultures
2. As barriers to buying satellite imagery come down, startups will gradually proliferate

I've written before about what kinds of companies I *don't* think people should be starting more of in the satellite imagery space (platforms & marketplaces).

So, to balance the scales, here's my personal "call for startups" in the satellite imagery space 🧵👇 1. Commercial roof reports

Seriously? Seriously. It's something like a 9-figure (!!!) business for @EagleViewTech auto-generating roof reports for commercial roofers bidding on jobs. Go do it in a new market or a better way.

1/7 I want to call attention to the National Geospatial Advisory Committee (NGAC) fgdc.gov/ngac. It's a group of volunteer experts who advise the federal gov't on policy regarding geospatial data and technology. They're seeking nominations for new members through Feb 22 2/7 The committee is mandated by congress through something called the Federal Advisory Committee Act (FACA) and meets four times a year in open sessions. They report to the Chair of the Federal Geographic Data Committee (FGDC) under the Department of Interior.

1/6 🧵 time! This is my second week at @umbraspace (umbra.space), and things are off to a quick start. The big news is that we raised a bunch of money, woohoo 🥳. But lemme tell you about what's more interesting: what we're gonna do with the 🧀spacenews.com/umbra-raises-3… 2/6 Buying satellite imagery today sucks. It's harder than buying a house. It usually takes months, and you almost always have to call in favors just to get anywhere. It's totally broken. If you follow me on here, you already know my feelings about it.

1/11

The most important deal of the last decade in earth observation: when Google sold Planet their Terra Bella constellation in 2017. So much to infer from that one transaction. Allow me to explain... spacenews.com/planet-confirm… 2/11

Some context: in 2014, Google acquired Skybox Imaging for $500M, the first “small sat” startup to achieve an exciting outcome for its investors. An avalanche of enthusiastic VC investment in the sector bookended the acquisition—it was a high water mark for the industry.

https://twitter.com/mouthofmorrison/status/1163558859342045184

Here’s a thread of 10 strongly held opinions I have formed about the “geospatial industry”:

1. The most successful and ambitious mapping project of all time, Google Maps, is an advertising platform. There is no “geospatial industry,” only industries with spatial problems. 2. It follows, then, that the most valuable geospatial applications are always custom-built in service of a particular domain that doesn’t self-identify as “geospatial.” The closest you can get to a “geospatial app” is a UI on top of a dev tool.

I’ve told this story in private a few times but it’s time to tell it publicly.

How I Got a Lifetime Ban from AWS: A Cautionary Tale for Idiots on Computers

(Illustrated with Larry David GIFs) The year was 2015, and I had just quit my first job out of college after less than a year. Not exactly the roaring start to my adult life I had hoped for.

I was pretty broke. But I had prospects—May of that year I got a job at @azavea and things were looking up!

1/ If you're curious about what the finances of running a world-class AI company looks like, you can get a sense by looking at IRS filings from @OpenAI, which we have available as recently as 2017. 2/ The first number that jumps out is that they spent over $28M in a single year with a headcount of just 99 people. And here I thought software was supposed to be cheap to make! Employee compensation alone accounted for $15.7M in expenses (over $13M of which was salaries).

1/ This is an interesting article from @philosophygeek: TL;DR It’s Descartes Labs rational for no longer selling data products (e.g. corn forecasts) and instead switching to a platform model so customers can build their own tools. Here’s why it’s scary. link.medium.com/VgLuVx33VV 2/ In my opinion, the “apply AI to commercial use cases” boom began in 2013 after Climate Corp was acquired and companies like SpaceKnow and Orbital Insight were founded by ex-engineering leaders from there. They proved you could raise money on the premise detailed in the post.