Intro
Today I came across a piece of advice from former NASA Astronaut and SpaceX Director of Space Operations Garrett Reisman where he gave a quote about SpaceX’s risk-taking failure paradigm which really stuck out to me:
We took very large risks when the consequences of failure were low, so that we could figure out all the things that can go wrong, fix them, so that when the consequences of failure are high—like when my friends are sitting on top of that rocket—the risk was very low.
Essentially, he said that it’s all about taking hundreds of “low-stakes” risks rather than putting all of your eggs in one basket and taking one high-stakes risk and hoping that everything works once everything’s put together.
With the SpaceX way (whether with Falcon or Dragon), as Reisman puts it: "the consequences of failure are minor, but the benefit is major."
I feel like this applies to a lot of areas of life and thought it would be worth its own short article since I think it translates to quite a few areas of our lives and careers.
A Few Examples
SpaceX Starship
The example that Reisman talks about in this specific interview is SpaceX’s starship.
Instead of waiting until humans are onboard to fly and test it for the first time, they instead fly dozens (or potentially hundreds) of uncrewed test flights before they fly it with people onboard.
They took a bunch of quick, successive big risks while the cost of failure was low, found out the root causes and learned from those quickly, and by the time it flies people, the system will likely be much safer once people are eventually onboard (once the cost of failure is “high”).
Career, College, Grad School Admissions
Imagine an engineering student who eventually wants to work in Guidance, Navigation, and Controls when they graduate in 4 years for example. The principle applies all the same.
If that student waits until the end of their four years to start taking things seriously - i.e. if they wait until they need a job - they’re putting all their eggs in one basket that they’ll get their job search right in this one high-leverage, high-stakes moment.
But if they didn’t do anything during their entire four years to build up their GNC portfolio, they’re probably going to catastrophically fail since they didn’t build in those “low stakes failures” throughout their four years of undergrad.
On the other hand, if you join a student group (e.g. USC’s Rocket Propulsion Lab) and work on GNC starting your freshman year, have a lot of “low-stakes” failures and incremental improvements, and apply to internships, interview for them, and fail (but learn from those failures) along the way, then you’ll probably be in a much better position at the end.
You’ll have figured out how the GNC world works, how to get a job, what types of jobs there are, and most importantly – you’ll have built up enough of a portfolio so that you have leverage to get recruited for top GNC jobs – so that by the end, you’ll be able to much more easily land a top GNC role.
I’ve personally seen this even at USC’s Viterbi School of Engineering – people who join student groups early, go through the learning curves and growing pains, but who eventually graduate straight into top jobs at companies like SpaceX, Blue Origin, Anduril, top startups, etc.
On the other side of the coin, I’ve seen people who – despite going to USC – did nothing outside of their coursework for their entire Bachelors or Masters, but who struggled immensely to find good jobs – or any job for that matter – after graduating. Some even took a staggering 6-12 months of applying to find any starter-level job, and this was despite going to USC.
The same could be said for both getting into (and getting a full scholarship for) college or grad school – you’re best served by incrementally improving from a lot of “low stakes” failures over time so that in high-stakes moments, you don’t endure long-term catastrophic failure.
It could be that you have a bunch of “low stakes failures” in terms of grinding away at practice problems or going above and beyond for studying for your classes, so that when you take final exams and midterms (which are high leverage moments that you’re graded on), you’re able to succeed without breaking a sweat.
Furthermore, another example is that it could also be about being a part of a student group or a research lab, where you do a bunch of experiments or build a lot of things which fail, but you learn from those failures, iterate quickly, and produce meaningful outputs which will move the needle for college or grad school admissions – such as publishing peer-reviewed research papers, building something really cool, or winning top competitions, etc.
Long story short, commit to a lot of low-stakes failures over time so that you don’t fail catastrophically at a high-leverage moment.
Content Creation
Another great and tangible example is content creation.
Imagine that, for some reason, someone invested $50,000-100,000 into a completely fresh YouTube channel and tasked you with making it successful.
Would it be better to invest all $50,000 or $100,000 into a single video and hope it just becomes successful, or to invest $1000 into 50-100 videos and incrementally improve over time?
Steven He – who I made a full-length article on after his visit to USC last November – actually went through a similar situation himself, and this was after his channel was ALREADY successful and had millions of subscribers.
That last point actually underscores the following point – Steven, a highly experienced multi-million subscriber YouTuber – actually failed miserably when he himself put all of his eggs in one basket.
Steven took over a year and a half off to shift away from his viral shorts to put $400,000 of his own money into a single longform production project which catastrophically failed and tanked the views on his channel and so on and so forth.
After he uploaded the project, his channel views tanked by over 80-90%, and he’d wasted an entire year and a half and nearly half a million dollars to learn this hard lesson. More details are given in the Steven Article I wrote last year.
The primary takeaway is that going forward, instead of making the blunder of going all-in on one thing again, Steven mentioned that he’d actually use the money to do something very similar to Reisman’s “Low-Stakes” risk focus rather than going all-in on one “high-stakes” risk.
Specifically, in an appearance on the “Driven” podcast a month later, Steven was asked “If you were given $100 Million dollars to make one video, what would you do with it?”
He mentioned that it directly conflicts with what he would do with it, in that he wouldn’t make one video if he had a $100M check due to the lessons that he learned from the previous show that he made which we talked about above.
He mentioned that, instead, he’d use the $100 Million to make AT LEAST 100 videos, and that he would bank on the optimization process which I summarized from his talk at USC.
The optimization process is that, over the course of making those videos, he would find enough problems in the videos and that he’d find the solution to all of them, so that by video 100, the system runs completely flawlessly.
Every problem they could face by video 100, they have faced, and could solve. He thinks that video 100 is a lot, and that he has no doubt that by video 100, he would have no doubt that it would work out.
Another quick example might be someone like Chris Stocks (Buchalo) who started doing interview-style videos when he was a freshman at Iowa State in 2022. By doing a bunch of low-stakes, fun interviews and by starting that early on in his college career, he was able to easily transition into full-time content creation by the time he graduated 4 years later.
Had he instead waited to start doing interviews until much later on, he might not have had the leverage to go full-time as soon (kind of like me at USC, in a way, lol), but since he’d taken a lot of low-stakes failures in an efficient way, he was able to go full-time by the time it mattered most.
Eric Suerez, Brady Your Tutor, and Casper Capital – who I highlighted in this previous article – are all great examples as well.
Overall, Steven’s framework is very similar to the SpaceX low-stakes vs. high-stakes methodology mentioned earlier, and pretty awesome to see that it holds up so beautifully between the two completely different domains.
Comedy & Soft Skills
I already mentioned this example in a previous article, but comedy (and soft skills in general, friendships, relationships, etc.) is a great example of this.
Imagine that there’s two aspiring comedians who are trying to get good at Comedy: Comedian “A” and Comedian “B” who are both new to comedy and aren’t that good when starting out.
They both take completely different approaches to comedy - Comedian A takes the approach of trying to get their comedy set right on the FIRST try.
Comedian B, on the other hand, just dives right into it. They do their first set and get no laughs.
They’re embarrassed, but they learn much quicker, and then they’re able to get some laughs their second time.
They do 10 more sets and most of their jokes bomb, but some actually start working super well and they keep repurposing and doubling down on them.
By the time Person A even attempts their first set, Person B has done 50 - and despite sucking badly for the first 20-30 sets, they’re actually pretty decent at comedy at this point.
Person A, on the other hand, just attempted their first set after Person B just finished their 50th.
After Person B finishes to a roaring applause, Person A comes on stage and gets zero laughs.
One person took imperfect action and converged to proficiency and mastery quickly, while the other didn’t.
Again, this is very similar to Reisman’s “lots of low-stakes vs one high-stakes failure” methodology, just extended to yet another domain.
In order to get better, you have to put the reps in, and most importantly, GET FEEDBACK on what worked and what didn’t so that you can improve between iterations.
The same thing even extends to dating, friendships, and other soft skills, etc. as well in much a similar way.
For example, with relationships, you first go on a lot of low-stakes dates (or put yourself in lower-stakes social situations in general) throughout your younger years before you do the high-stakes thing, which is getting married.
In modern day society, lower-stakes things such as going on dates (or asking people on dates in the first place) – the equivalent of the dozens of uncrewed Starship test flights that you learn from – help mitigate the risk before “strapping yourself into the rocket” of marriage if you go about it correctly.
The former (if done respectfully) may still seem “risky” in the moment, but ultimately are still low-stakes and will lead to a much lower-risk outcome when you go into the high-stakes thing.
Feedback is Crucial
Given all of the above examples, one critical element to highlight is the importance of feedback.
You definitely need to put in the reps, fail fast, and iterate quickly with this methodology, but in order to progress as fast as possible and make meaningful progress, you must get the right kind of feedback so that you know where to improve.
With SpaceX, it’s about having the right kind of sensors and telemetry so that you can diagnose where the mission went wrong if you did a flight with Starship for example – was it one of the engines? Was it a random valve? Was it some other part of the system which you didn’t expect to go wrong? You need to know this in order to fix it and improve it next time you fly.
Similarly, for the comedian, performing and getting reps in is important, but if you perform to an empty room, that will be a lot less effective than performing to actual audiences and getting the real instant feedback (and receiving the pain of failure to motivate you going forward).
Comedian Steve Martin – who relentlessly worked for 14 years at honing his comedy skillset before becoming a breakout success in year 15 – is one of my biggest inspirations when it comes to career improvement, and he occasionally did perform to empty rooms, but for the vast majority of the time, he had some sort of audience to give him that feedback.
With regards to content creation, I also came across a very similar insight recently.
I’d previously found some success with filming videos in my room filming by myself for the first few years of being a creator (although there were some long periods where I created nothing due to obligations, money, time, other excuses, etc.).
This summer I took the plunge – inspired by Chris Buchalo, Eric Suerez, etc. – and started doing campus interview style videos for my content – something I’ll definitely do a lot more of going forward (whether on campuses or in public in general).
The main insight I gained from that is that the people in my videos were essentially an extension of my audience and I got to see their reactions real-time.
I got to see what they reacted well to, what didn’t work as well, and what worked live – very similar to a comedian performing to a live audience versus performing in an empty theater.
Even if it still ends up being a long process of failure + feedback + iteration + improvements before I get there, I’m confident that if I continue doing interview style content going forward, I’ll improve rapidly and be successful.
With interviews, not only do I get good feedback, but the feedback loop is so much faster, and combined with a form of the “optimization process,” it’ll make me successful much faster.
Optimization Process
One last thing beyond feedback that’s important is to incorporate some degree of first-principles (as much as you can), Steven He’s optimization process, or both if you can.
For example, with Rockets, understanding the underlying fluid mechanics, liquid rocket propulsion, gnc, aerodynamics, and materials science will help you solve the problem much quicker and more efficiently than if you didn’t understand those things and just kept changing stuff until something stuck.
With things such as soft skills, it’s slightly different but still very similar.
With comedy for example, when I was improving, putting sets in and getting feedback from those sets was critical, but concurrently I was also studying the comedy principles from Joe Toplyn’s 2014 book “Comedy Writing For Late Night TV” (specifically, chapters 1-4 and 5-6).
The principles that Toplyn outlines in his book – especially in chapters 5-6 – are the equivalent of “first principles” for making good jokes. While I could’ve probably still gotten good at comedy just on repetition and feedback alone, reading about and implementing the concepts from this book while concurrently doing those things exponentially sped up my progress.
It’s also sort of similar to AI and Machine Learning - of which I’ve taken 20+ credits at USC as electives. While I’m vastly simplifying things, AI learns best from statistically learning from A TON of examples, and in some cases can also benefit from incorporating additional rules, constraints, or domain knowledge.
The human brain works much the same way – if you can train it on a lot of examples, repetition, and first-principles, then that’s where learning is optimized.
First principles - and "always understanding the why. Always aiming to understand the why" behind things (as Reisman's former PhD student always drilled to me!) - is one of the best ways to solidify information.
If you can't figure out first principles, then sticking to something similar to Steven He’s optimization process (for fields such as content creation for example, where there aren’t necessarily established first principles like engineering or rocketry) covers the first two of those three learning principles and is a good framework to use.
(After all, AI can be trained essentially the same way - they can be trained and tuned specifically based on examples only, and produce decent outputs)
Conclusion + Summary
This is the only section that I’ll write using AI (I don’t write my articles with AI at all), but I’ll use it since it can re-articulate everything in a different way than I can and repeat the most important lessons more eloquently.
The main point is to not try to avoid failure, but to instead structure your life and career so that failure happens early, rapidly, and repeatedly in a low-stakes way – before the stakes become enormous in life.
Garrett Reisman’s quote from his news interview which sums this up nicely was:
We took very large risks when the consequences of failure were low, so that we could figure out all the things that could go wrong, and fix them, so that when the consequences of failure are high—like when my friends are sitting on top of that rocket—the risk was very low.
Reisman’s SpaceX analogy of essentially “flying the rocket 100 times, breaking it, figuring out what went wrong, fixing it, and flying it again before eventually putting humans onboard” is probably the best way to illustrate the entire concept.
This high-risk, low-stakes, rapid iteration, rapid-feedback framework is one of the best ways to make progress in many areas of life and also allows you to make that rapid progress without also putting all of your eggs in one basket and banking on a single high-leverage moment.
Whether it’s SpaceX flying humans into space, a student looking for their first job or studying for a final exam, a comedian looking to get better at comedy, or a creator making 100 videos and iterating instead of banking it all on one, the underlying process is the same.
Spread your risk out across a lot of low-stakes failures, and learn & iterate quickly between those, so that when the high-stakes moment comes, the risk is low.
That's how you maximize your chances of success.