Testing is Like Finding Exoplanets

Astronomers have been finding lots of planets around other stars, which have come to collectively be called exoplanets. And, as part of that endeavor, they also try to think about finding life on those planets. There’s lots of corollaries here in terms of thinking about testing.

Before I get into the corollaries, I want to remind readers that I’m generally against arguments that are ostensibly supportive of testing but often have little to no impact on how testing is conceptualized as a fun and distinct discipline. Sometimes far from having little to no impact, it’s even worse: certain arguments may be having a detrimental impact on how testing, and those who practice it as a specialty, are perceived.

I will also remind readers than when I get into this frame of thought, I’m thinking along the lines of Testing is Like… which I want to encourage more testers to do.

Okay, so exoplanets. What am I actually talking about here?

Articulating Why Something is Difficult

First, one thing that it helps to do is be able to explain why your discipline is complex and/or complicated. I actually find that testers struggle with this more than they should. (I tried my hand at why I feel testing is complex and why testing is complicated.)

So let’s practice. You don’t have to be an astronomer to guess what makes finding exoplanets hard. So take a second before reading on and think about that. What makes this task difficult?

Have you thought about it?

Seriously. Take a second to practice this. It’s instructive to be able to articulate, down to details, why a particular discipline or specialty is difficult.

The most obvious answer here is that, when looking for exoplanets, you’re looking for relatively small objects that are very far away. Yes, of course, the objects — being planets — are large in actuality. But they are very small relative to our distance from them.

But it’s not just that, right? What else?

Stars generate their own light but planets don’t. Planets shine only by reflection. So the problem is actually that we’re looking for relatively small objects that are very far away and quite hard to see.

And there’s yet another problem. A planet is going to be relatively close to its parent star. (Again, from a human perspective it may be very far away from it. But at the distances we look for exoplanets, it’s practically right on top of it.) Further, that star is an extremely bright light source. As you can no doubt imagine, separating out the light reflected from a planet from the glare of its parent star is particularly difficult.

So the problem is actually that we’re looking for relatively small objects that are very far away, are quite dim, and are next to something that is very bright.

Difficulty Encourages Techniques

Okay, so it’s hard. But how do we do it?

One way try to observe the effects the planets have on their parent stars. Basically what this means is that you would be looking for a certain “stellar wobble” in the path that the star takes through space where the “wobble” is caused by an orbiting planet. Obviously that gets a bit tricky and error-prone because stars are so much heavier than their orbiting planets.

Another way is referred to as “radial velocity measurement.” The idea here is that as the star and its planet move around their center of mass, the star moves toward us for half of the revolution and away from us for the other half. What this would mean is that the light we see from the star is red shifted for a while and then blue shifted. (Essentially the Doppler effect.) So this technique measures the motion of a star toward and away from us.

This is actually still looking for the stellar wobble; it’s just using a more refined technique. Rather than looking directly for the stellar wobble, you look for it indirectly.

That Can’t Be There!

Now let’s just consider this: “Oh, there can’t possibly be a bug there.”

Surely most of us have heard some variation of that, right?

Astronomers, prior to 1992, would have said something similar. “Oh, there can’t possibly be a planet there.”

In this case “there” refers to the orbit of a pulsar. Pulsars are basically small, incredibly dense, rapidly rotating masses of matter that are left behind when a large star explodes in a supernova. This explosion essentially blows huge amounts of stellar material out into space. Astronomers and astrophysicists would (and did!) expect that any planet unfortunate enough to be in orbit around such a star would be completely destroyed.

Yet, in 1992, we found a planet that is orbiting a pulsar that’s referred to by the catchy name PSR 1257+12, which is about 2,300 light years away from us. In fact, not just one planet: but two planets appear to be orbiting this stellar remnant!

So, as in our testing careers, we often have to be prepared to dispel illusions. Here the techniques we used were those developed by humans but then augmented by tooling that helped us observe the planets and then perform calculations on what was being observed. See how much more fun this kind of discussion is than discussing “testing” versus “checking” and harping on about how “automation isn’t testing”?

Well, I think it’s more fun, anyway. But let’s continue…

Be Careful of Standards

There was a concept that astrophysicists had come to call the Goldilocks planet: not too far, not too close, not too hot, not too cold, and so on. Basically “just right,” as in the story of Goldilocks and the Three Bears. This notion of “just right” was effectively based on our knowledge of our own solar system where we basically had a sample size of one: life on Earth.

Yes, there was a lot of life but it was life based on a common chemical background located on just one planet. So an assumption formed that a planet like ours in most respects should be what we search for. This would be akin to assuming that how we use an application is how all users will use the application and thus we should only concern ourselves with that “happy path.”

So we started out hunting for elements along this happy path. And we pretty quickly found that our standard was wrong — or at least extremely misleading — and our happy path was actually probably not all that common or prevalent.

The Happy Path Had Edge Cases

The technique originally available for exoplanet detection involved, as I said above, measuring the small motion of the star that could be determined as occurring due to the gravitational pull of a planet orbiting it. The problem was that such a technique was likely going to be best at detecting large planets, where “large” means those planets capable of exerting strong gravitational pulls on their parent star.

But let’s put ourselves in someone else’s shoes, which is what testing often requires us to do. In this context, let’s put ourselves in the shoes of someone observing our own solar system using this technique. Such a person would see the effects of Jupiter before seeing the effects of Earth. In fact, they might not even know that Earth is there.

In any case, when this technique was used to search for exoplanets, the first positive results were the discovery of what came to be called “hot Jupiters.” These are massive planets — typically several times larger than our own Jupiter — that are orbiting close to their parent stars. Very close. In fact, often closer to their parent star than Mercury is to our own Sun.

Once again, we sort of had a situation of “A bug can’t be there” but this time it was more like “That type of bug can’t be there.” A large Jupiter-sized planet as close to the Sun as Mercury? Ridiculous! Can’t happen!

But then something interesting occurred. We started to find only hot Jupiters. This would be akin, perhaps, to finding only certain kinds of bugs during our testing but literally no other kind of bug at all. Now, to be sure, it wasn’t that our observations were wrong. We were, in fact, finding hot Jupiters.

But we were finding so many that it seemed to call into question how our own solar system formed. Our own solar system started to appear quite unique. But here it’s the nature of the testing we have to question. Case in point, with the astronomy example, the fact that we were finding hot Jupiters first was simply a result of the detection system available. We were finding the equivalent of “shallow bugs” due to the nature of our observation technique.

Then we changed our search ability. Starting around 2009, we were able to search for the small dimming of a star’s light that occurred due to the passage of a planet across the visible face of the star. The idea was encapsulated in looking for the “light curve.”

This is essentially looking for “dips” in the starlight as the planet temporarily moves in front of the star.

Using a Different Test Technique

Thus was a new testing technique devised. However, like most techniques, it did have a constraint built into it. This new type of search (“test for exoplanet”) would be successful only if the orbit of the planet is oriented so that the planet passes between its star and Earth. A planet whose orbital plane is perpendicular to that line of sight would effectively be invisible to the technique and thus to the test.

So, a great test — but it had detection limits and thus it had a selection effect of what it was likely to find.

Now, also of note, this testing was done with the help of tooling. In this case, the Kepler satellite. But that satellite searched only a small segment of the sky; effectively only an area that was a couple of times bigger than the full moon. Consider:

That’s a pretty limited range of test applicability, even just in our local space. Yet, on average, astronomers were now discovering exoplanets at a rate of about three per day. But still: this is very limited in scope. Consider:

What that basically shows you is that the Kepler satellite has a test coverage of only about 100,000 stars. Yes, that sounds like a lot but that’s only about 0.2 millionths of the number of stars in our galaxy alone.

So while the test technique had a way to remove the selection effect of tending to find hot Jupiters, it had a constraint that made it difficult to apply, and also relied on tooling that was not as comprehensive as we would have liked.

Any specialist tester can relate to that!

More Edge Cases Appeared

We eventually started to discover many odd types of planet out there.

For example, in 2012, the planet called “55 Cancri e,” located around 40 light years from Earth, was discovered. The planet appears to contain more carbon than our own Sun. And from various forms of analysis, astrophysicists have determined that there could be a thick layer — by “thick”, I mean a third of the radius of the planet — made up mostly of diamond. At greater depths the diamond could possibly be in liquid form, which takes us into new areas of chemistry.

And the idea of a new chemistry gives us another way to start thinking about exoplanets: life that might exist on them.

Hunting for Life

The idea of life out there in the universe has one that has long fascinated and now that we’re able to discover planets, the idea of using our test techniques to see if life might exist on those planets is more in reach. In that context, we have to focus on a few things.

  1. What do we mean by “life”?
  2. What do we know about how life could start (origin)?
  3. What do we know about how life could change (evolution)?
  4. How can we detect life?

That last one is really crucial. For example, let’s take our own solar system. There are some exobiologists who believe that perhaps life could have formed on Titan, which is a large moon of Saturn. Titan has extremely low temperatures, however, which means chemical reactions on its surface take place extremely slowly. Any organic chemical processes that take place on Titan would take place very slowly. In fact, they might be so slow that don’t recognize them as life.

And what about that planet with the possible liquid diamond? Would we even recognize a form of life in that context, even if we were staring right at it?

Applying Imagination

At this point, we move into a mode of thought that requires a lot of imagination as well as some knowledge of basic science. Total side note here: I once wrote a paper called Truly Alien where I tried my hand at applying such imagination.

In terms of applying our imagination, we look at typical exoplanets and try to figure out how the basic rules that govern the development of life would operate in the environment of each one. We ask how, where, and which sorts of life might develop on those exoplanets, and then we speculate about how some sort of civilization of life might come into existence there.

So think of our “bugs” in our applications as a life form. It’s one we are seeking to find. Some of these will be simple bugs (akin to perhaps single-celled organisms) and some may be very complex, operating due to the intersection of technology abstractions (akin, perhaps, to an advanced technological civilization).

Thus what a lot of what testing comes down to is the hunt for bugs but also the hunt for where they might live. And where they might live gets into understanding the “ecosystem” of such life.

Consider that much of the history of life on Earth depends on the details of the environment in which those steps played out: the specific setting of our own planet. For example, that Goldilocks planet concept that I mentioned before was predicated upon an assumption which was that “just right” meant a planet that formed far enough from the Sun (but not too far!) such that a large body of liquid water would exist.

But is that necessary for any form of life? So our question, then, becomes: how would these steps for the development of life work out in the kinds of radically different environments we see on exoplanets? Would life develop there as it has on Earth? Would it develop differently? How different could it be? What kinds of life can we imagine in the newly discovered area of exoplanets?

Think about this in relation to how testing is often put in service to understanding requirements on a project, where there are often numerous user stories competing for attention, all of which may eventually “evolve” into some kind of feature that takes in input and produces some output. We essentially have an ecosystem!

Boundaries and Edges

All of this means we can think of life in a few ways:

  • Life that is like us. Meaning, life based on the chemistry of molecules containing carbon atoms.
  • Life that is not like us. Meaning, life that is based on chemistry, but not necessarily the chemistry of molecules containing carbon atoms.
  • Life that is not even remotely like us. Meaning, perhaps, life that is not based on chemistry at all.

What we end up with here is classes of things to investigate and some of those may be equivalence classes. All of these have boundary conditions at which we will consider edge cases. Certainly testers can see where I’m going with that, right?

Obviously, dealing with questions such as these must involve a great deal of imaginative thinking. Nevertheless, there are some basic laws of nature that operate everywhere in the universe (so far as we know, anyway). Further, these laws set limits — albeit often generous ones — on the ways that we can think about life in other places or in other conditions. And, to be sure, there a large number of wildly different scenarios we can imagine playing out even within the limits set by those laws.

Certainly the corollary here with testing is front-and-center, yes? There are certain “laws” that dictate how applications or services work and, for the most part, they operate anywhere that electrons within a computer are constrained to form certain 0’s and 1’s that (ultimately) make up the chemistry of our applications.

But far from meaning that testing these different “configurations” is easy, it means there are wildly different scenarios that we often have to account for, even given the fact that everything reduces to utter simplicity.

Biases and Chauvinisms

Thinking about exoplanets, and particularly about the nature of life on those planets, is full of what are termed chauvinisms. The most common of these is referred to as “carbon chauvinism.” I alluded to this earlier. This is the notion that life elsewhere must be based on carbon, since that’s what our form of life is based on. And carbon chemistry, which presupposes molecular chemistry, proceeds most quickly and efficiently in liquid water. Hence the desire for planets that have large bodies of liquid water.

There are, however, two other chauvinisms that can be a bit more subtle. One is referred to as “surface chauvinism.” This basically refers to the idea that life has to exist on the surface of planets, which is really just a sub-category of what the science-fiction writer Isaac Asimov referred to as “planetary chauvinism.”

This is actually an odd chauvinism that still asserts itself and I say odd because we know of numerous examples of subsurface microbial life on Earth. But what about something like Titan, the moon I mentioned earlier, where there are clouds of ethane and methane that interchange material in complex ways with what seem to be lakes of liquid ethane or methane. Could that lead to some form of life that we currently don’t even think about?

Another example is what is referred to as “stellar chauvinism.” The idea here is that of an assumption that any planet which could support life must be orbiting around stars. Yet as recent tests and observations have started to indicate, it’s likely that so-called “rogue planets” or “unbound planets” may actually be more common than the traditional concept of planets that orbit stars.

And consider here all those techniques I mentioned above. They wouldn’t work to find these kinds of planets at all! So how are they found? Well, that has to do with a technique that observes gravitational microlensing events. What’s interesting is that these are very rare because they require the precise alignment of some background star, a (relatively) massive foreground object and Earth.

Back in 2011, there was the discovery of 10 short-duration events of this type over the course of two years. That doesn’t sound like much, right? But when you consider the size and distance scales involved and the relative rarity of such events, this actually suggests there is a huge population of these rogue exoplanets throughout the galaxy.

But notice how much time and effort it would take us to remove the stellar chauvinism compared to, say, the surface chauvinism.

This is often the case in testing: the harder things to find require a greater outlay of effort, particularly when you are trying to dispel illusions about “how things work.”

So the overall point here is that astronomers and astrophysicists could use these biases and chauvinisms as a starting point to ask about what others chauvinisms might exist. How do we recognize them? How do we combat them? And that should get testers thinking about what kind of “test chauvinisms” exist.

The Fun of Other Disciplines

I hope this article was interesting in its own right but, more importantly, I think having discussions like this — as well as bringing in other disciplines that also use testing as a core of their approach — is a very viable mode of getting people across various different industries interested in testing and understanding why it requires human thought and action that can be supported by (but never replaced with) tools.

This is more important than ever when testing is often marginalized or conflated with other roles or, even worse, relegated solely to the use of tools.

However it’s because those very things happen that I feel we, as test specialists, need to make sure our articulation of the testing discipline is not just accurate but also indicative of how interesting testing as a whole is and how it is like so very many other endeavors of human thought that require creativity, imagination, exploration, investigation and discovery.

Just as various areas of science, like astronomy and astrophysics, have had to work on their narratives to keep the public engaged and interested, I think us test specialists have to do the the same thing. I’ve previously indicated my view that testers should avoid being such testers” and part of that is working on our narrative intuition a bit, which is something I’m continually striving to do.

Hopefully this article goes some way not to just showing you what I consider to be the interesting corollaries between testing and another discipline but also to recognize the broader, fascinating universe that we are part of. And being able to consistently feel wonder at the sense of discovery is one of the hallmarks of testers everywhere!


This article was written by Jeff Nyman

Anything I put here is an approximation of the truth. You're getting a particular view of myself ... and it's the view I'm choosing to present to you. If you've never met me before in person, please realize I'm not the same in person as I am in writing. That's because I can only put part of myself down into words. If you have met me before in person then I'd ask you to consider that the view you've formed that way and the view you come to by reading what I say here may, in fact, both be true. I'd advise that you not automatically discard either viewpoint when they conflict or accept either as truth when they agree.

One thought on “Testing is Like Finding Exoplanets”

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.