Entrepreneur Introspection
Circles of trust and competence
For those that haven’t been following along, I have been using this forum over the last few years to document the development of Automunge, an open source platform for preparing dataframes for machine learning. We have come a long way in that time. That is an understatement. However one of my fears for this project is that we may meet all of our goals for product quality and functionality and yet still fall short attracting reputable users at scale due to what increasingly appears to be a credibility gap for me, the solo founder (slash author / architect / developer).
On paper my resume is lacking what may be a few expected conventions:
- I don’t have a phd
- I don’t have a cofounder (yet)
- This is my first software startup
- My prior professional background is mostly in unrelated fields
I have tried to avoid making the startup, and the book of essays for that matter, about me. The software and application are much more important than any quirks of the founder. The reason for this essay is the growing recognition that others might not carry that same attitude, so am trying to get ahead of any diligence questions that may arise. I have been blogging for six years now and this will be the first essay addressing biographical aspects of the author (perhaps with the exception of an essay about Christianity a while back), so I hope a reader may grant a little leeway for the indulgence.
Probably my most impressive professional credentials include serving as marketing lead for successful execution of over $1B in contracts in the wind turbine supply chain, including overseeing contract negotiations and managing commercial strategy for these sales. The funny thing is that I had done much much more business with big tech firms before joining the field of data science than in the time since. Prior to wind turbines I studied engineering at the University of Florida, and have degrees in mechanical and systems engineering as well as an MBA. Automunge isn’t my first startup, although is definitely the furthest progressed. I experimented with university technology transfer in 2007, which was a learning experience. I started exploring forming a residential solar startup in 2017 and ended up pivoting to a salary position. Some of the diligence for the solar venture made it into Book 1 of the essay collection.
More recently I had a not insignificant amount of professional challenges originating from difficulty maintaining traditional working hours due to sporadic sleep issues. I now know that a big contributor was originating from symptoms surrounding extensive food intolerances, which identification was quite a challenge due to at times chaotic surrounding symptoms and delayed responses to food intake. It took years. Part of my motivation for publishing essays and developing software is that within this period I have also had difficulties with verbal communication and vision impairment, which I now know was a byproduct of symptoms from food intolerances as they mostly resolved once getting my diet figured out. I was trying to find other ways to contribute. Some aspects of my personal life have suffered from complications of health and nontraditional professional endeavors, I have always had the hope that short term sacrifice of focus on building something could open the door to other things down the road. Have a great family though.
When these medical issues began interfering with my ability to meet regular working hours I was faced with a choice. I could have sought to draw from disability or became litigious. Instead I devoted full focus to building something from scratch. It has not been an easy path.
Along the way what started as a writing project has transitioned to a building project and more recently a channel for research. That being said, research for research’s sake is what academia is for. I’ve been more interested in research as a channel to support entrepreneurial pursuits. The particular model I am following aligns with common academic conventions such as public disclosure. I have consistently found that writing papers helps to identify gaps and potential extensions for the software, and building features similarly fuels research. They are mutually reinforcing.
Over these last three years of full time focus I have made rapid progress in both software and research. I haven’t ruled out pivoting to a salary at some point but even if make that decision expect would be in a much better position to negotiate having met a few thresholds with the startup. Applying for role prior would be foolish. This software startup has been my sole focus for most of that time. Part of what has motivated me is that I have increasingly become convinced that the platform may become beneficial for researchers targeting aspects of pandemic response. This is why I have been working so hard and why I continue to do so.
I have also discovered something important about myself along the way. I am simply a much better communicator in writing. Independent of health issues, it is like I have a more direct conduit to my thoughts that way. At times I can actually communicate faster in writing than verbally. It is both a blessing and a curse. It has been interesting to see this become aligned as I have become a more proficient coder. One thing I’ve noticed is that the more time I’ve spent coding there have been changes to how I read books. When you are viewing code it’s a different kind of reading. You take in pages in aggregate, grabbing on to patterns and cherry picking points of relevance. It is much less linear. I think it has changed some of my patterns of thought in conjunction. I practically have the Automunge codebase memorized at this point.
Full disclosure: I did not start out as a proficient coder. In fact, the Automunge project in the earliest days was more of a learning and building experiment than any traditional software venture. It took several months of experiments, essays, and iterations before the idea of creating a venture arose, and even then a side-track through the 9–5 world diverted the project for a few months before picking it up again.
It is hard to describe the exact path followed from learning how to code to attempting to start a business, so I’ll try to use some visual aid to demonstrate. To borrow the metaphor from the MBA tome Blue Ocean Strategy, consider that an industry’s competitive landscape is like a vast ocean, and within that landscape are populated islands of commercial offerings, in many cases adjacent or overlapping. Just like how car dealerships tend to locate next door to their competition, it makes sales and marketing easier when customers may be educated by your competitors if not by you, which is a common practice independent of industry. For this image, let’s say the blue region represents landscape of potential viable commercial offerings, and green regions what is offered by the market. It is an intentional framing that much of the blue region is absent of commercial offerings, after all there are still many discoveries to be made, inventions to be created.
Now consider each of these competitors has their own capabilities, talents, and expertise. Let’s call the regions that they would be capable of providing commercial offerings given sufficient time and energy (barring any competitor barriers) as their circle of competence. We’ll imagine that these firms are big tech, so we can expect that a circle of competence would be pretty large, potentially spanning multiple industries.
Now consider my position at the start of coding. There was no realistic chance of offering any form of commercial product, I could barely code. Thus my circle of competence was tiny and initial iterations were well outside of any viable channels of commercialization. Hence the air quotes around the word viable in Minimum “Viable” Product. It was a product, and it did what it was designed to do, but there was no realistic channel for commercialization. I didn’t have a big enough circle of competence to approach any of the viable regions, at least from a software standpoint.
So a great deal of the effort for at least the first year or two was iterating my way through continuous software development to expand the circle of competence. It would be hard to pinpoint exactly, but at some point my competence actually started to overlap with the space of viable commercialization channels. Most people have an inclination to wait to start building something until their circle of competence overlaps with an ocean. I am advocating that it is better to start building much earlier as the simple act of building greatly accelerates the pace of circle expansion.
One key though, and it is the lesson of Blue Ocean Strategy, it is not enough to reach the ocean, ideally you want to find an unpopulated region non-overlapping with your competition. After all why would someone source from a startup with limited resources when they could go to dependable big tech? You need a differentiator.
As a startup it is not enough to breach the viable region. That will get you revenue, it will get you started, but you will soon attract the competition. After all that whole time you’ve been iterating to competence they have been too. They will see you, and if you have success they will follow you. In the words of Charlie Munger, you need a moat. Realistically, you are not likely to build a defensible moat with a minimum commercial product. Expanding the abstraction a bit, you need to keep building your circle of competence until you’re further into the blue water territory with a more advanced region of the viable commercial space.
And this is where the Automunge approach has come into play. These last few years, the whole time that we have been expanding our circle of competence through software iterations, in parallel we have been establishing outposts of knowledge by research deep within the regions of viability. We’ve been reading papers, writing papers, attending conferences, and attending workshops — all at the cutting edge of state of the art. Artificial intelligence. Deep learning. Quantum computing. Quantum information. Consistently. Over and over. And it has paid off by giving us insights into regions well outside of what competence would have been accessible from software iterations alone.
The point, if I have one, is I think the Automunge project has started to reach the region where we are in sight of a long term viable business. And we are doing it with open source software. And a solo founder. And a dream. And although you might say that our heads have been somewhat in the clouds at times, I assure you that our feet are firmly planted on the ground. We have 60,000 lines of novelty to demonstrate. And we’re just getting started.
The field of machine learning has rightfully hesitated to marry between formal software practices and learning pipelines owing to rapid rate of change in the scope. We believe we have identified a segment of sufficient stability meriting formal standardization.
The field would benefit from less researchers and more engineers, and pretty soon we’re going to need a few. We’re looking for domain expertise in machine learning and quantum computing. We’re looking for coders. We’re looking for architects. We’re looking for communicators. And yes, we might even eventually pick up an MBA or two along the way. But first, we’re looking for a seed investor.
When Say coined the term entrepreneur (250) years ago, it was intended as a manifesto and declaration of dissent: the entrepreneur upsets and disorganizes. His task is “creative destruction”.
- Peter Drucker
Books that were referenced here or otherwise inspired this post:
Blue Ocean Strategy — W. Chan Kim and Renée Mauborgne
Innovation and Entrepreneurship — Peter Drucker
As an Amazon Associated I earn from qualifying purchases.
For further readings please check out the Table of Contents, Book Recommendations, and Music Recommendations. For more on Automunge: automunge.com
