Contrarian Beliefs of Asher Bond
Contrarians can get really along great. Who knew? Hit my line if you might be a contrarian too.
Peter Thiel is a leader of many venture thoughts and other big thoughts, not just contrarianism. I like to pick on Peter because he said it so concisely: “What important truth do very few people agree with you on?” as one of his favorite questions to ask people when getting to know them. He has a Book Zero to One if you want more context on that. To be honest I haven’t read the book, but that doesn’t mean you shouldn’t. Contrarianism is evident in Peter’s investments and investment thesis.
Asher Bond Ventures a contrarian fund too. I mean sure, for the moment there’s a smaller number of assets under management. My money’s small but my money’s smart. And the contrarian conviction is strong. Contrarian conviction is the type of fire in the belly that facilitates a proper run through the wall of popular convention to reach the truth. I thought I should go into what my contrarian beliefs are as a way of hinting at the general Asher Bond Ventures Investment Thesis and as a general way for people who read my blog to get to know me.
People can change their minds, but here are the important truths very few people agree with me on (for now):
#1: Wall Street is the Alternative Investment, not the other way around: I used to listen to “Alternative Rock” in the 90s and there was a compilation album I liked called No Alternative. It was full of so-called “alternative” rock songs that had at that time been flooding the mainstream airwaves and record stores. In the context of high alpha investing (for accredited investors), Wall Street alpha is the alternative to alpha coming out of Asher Bond Ventures, Founders Fund, VUVP Fund, (insert any fundamental early stage venture fund here). Getting in early and adding value is the fundamental way of doing investing but it’s hard and takes sophistication. There’s a convenient alternative for those who lack the grit, conviction, sophistication, accountability, discipline, capital, etc: Wall Street. That’s just one alternative. Wall Street is the alternative to me. I’m not the alternative to Wall Street. Does it sound like I have some type of chip on my shoulder? Hail-yeah it does!
#2: I need to see the chip on your shoulder to know that you have what it takes to do venture. It’s a popular belief that a chip on the shoulder is bad for workers to have. But founders are some of the hardest workers I know and I insist to inspect chips on shoulders to know people are doing the right things for the right reasons.
#3: The best are not always the brightest / the brightest are not always the best. Not everything is a spelling bee or a math competition. Also, I’m not that smart and I could win spelling bees. Often the best person is the person who’s most solid in terms of following through and staying committed and loyal, but who holds strong opinions loosely and may even take a naive and open-minded approach. If I could, I’d trade the smartest linear thinking brain for a brain that flawfully but eventually solves dynamic problems consistently. Show me the team that consistently solves the prisoner’s dilemma. I’ll be more impressed than I would be with the team who consistently solves the most advanced leetcode or math problem. Check out my leetcode profile for an LRU cache that will blow your freaking mind with performance, though. Turns out these type of caching mechanisms are helpful for AI & machine learning.
#4: AI is capable of reasoning / higher order functionality. Computers can almost instantly reason better than any of the of the brightest human minds when given specific problems that are well suited for statistical inference, machine learning, etc. In specific areas such as pattern recognition, optimization problems, or large-scale data analysis, computers generally outperform humans in both speed and accuracy. Reasoning that requires analysis of large known datasets can be done faster and more effectively with an LLM than without. And sure, we know that LLMs don’t do everything great and we don’t do everything in AI with an LLM because there are so many other tools.
#5: Programming is the least important Software Engineering function. Large Language models are better at languages (especially programming languages) than people are. Engineers originally built and operated catapults, so in a venture or launch sense software engineers still kind of are building catapults (or potato-launcher cannons, as seen on the mostly-ficticious TV show Silicon Valley.) The engineering re-catapultization is more and more prevalent these days since the advent of natural language understanding and code generation co-pilots. Generative AI is literally built for writing code. Let’s be honest, everyone’s doing it or trying to. Engineers are telling computers to write code and doing most of the planning by hand. There are many startups and incumbents trying to make AI do the planning too! A lot of people seem to want an autonomous software engineer agent, but they just can’t get the software design and planning to come together with the verification yet. Engineering is still hard. Not saying computers can never do it. They can do better than the so-called “engineers” who did nothing but generate code.
#6: Non-maintainable code should be written, executed, and put into production so long as it facilitates and maintains the intended user-experience. Maintainable code will be thrown away if it’s easier to integrate new code. New code is easy to generate. Believe everything I say and test everything. Trust but verify. Get to a level of conviction about your test harnesses instead of becoming overconfident about code culture. This is done by embracing behavior driven development or test-driven development. Writing software tests before writing application code is something the software industry pays a lot of lip-service to. It used to be that writing the test first required significant discipline and took a while to show increased velocity, but now with generative AI not generating the perfect response (being code that behaves properly) the first time we’re seeing a huge velocity increase right out of the gate for Test-driven DevOps approaches (which I’ve been screaming about and practicing since the 2010s).
