Updates
I'm hoping the holidays give me more time to explore some concepts I've been wrestling with related to financial projections and valuation. More details to come.
Thanks everyone who sent me links and recommendations this week. If you stumble across something interesting, send it my way! You can send any recommendations to links@houstonp.com.
Books
Seven Brief Lessons on Physics by Carlo Rovelli - I enjoyed Carlo's book on time, and this was more of the same. If you're interested in a shorter book discussing similar concepts, this is it.
Moonwalking with Einstein by Joshua Foer - A book on memory and memorization – this was a great read. There were tons of actionable insights that you could apply to your own life. I still remember the first example of a memory palace that he walks through. Who would have thought a story of competitive memorizing could be so interesting.
Time Travel by James Gleick - I only got through a third of this book. It was not what I was expecting. Perhaps it gets better, but it was a slog. There's not enough time to waste on things that bore you. Avoid.
Podcasts
Some extraordinary episodes from this week:
Kevin Systrom and Mike Krieger – How to Build a Great Product – Invest Like the Best
Darryl “DMC” McDaniels Finds Therapy More Helpful Than Olde English 800
I've been making my way through Rework's backlog of episodes:
Videos
Functional Composition - Composition is a powerful method used in functional programming. This is a wonderful talk using music to explain the concepts.
Articles
How to recognize AI snake oil - Enough said. There are people selling snake oil to the C-suite and the masses. It's dangerous and unethical. More people need to call out this bullshit.
AI today and tomorrow is mostly about curve fitting, not intelligence - Fancy curve fitting isn't intelligence. Stop trying to sell it for anything more than it is.
Are Neural Networks About to Reinvent Physics? - Hint: no.
Statistics vs Machine Learning - I'm very disappointed this article is on the Institute and Faculty of Actuaries website. As I noted above, machine learning is fancy curve fitting, i.e., statistics. Nothing about “the solution” stated in the article is specific to machine learning. Articles like this are exactly why I get frustrated by “machine learning” talk.
The Robust Beauty of Improper Linear Models in Decision Making - The power of a basic linear model should not be understated.
This article presents evidence that even such improper linear models are superior to clinical intuition when predicting a numerical criterion from numerical predictors.
Hidden Technical Debt in Machine Learning Systems - Even if you just read the abstract, this is important for decision makers to understand.
Machine learning offers a fantastically powerful toolkit for building useful complex prediction systems quickly. This paper argues it is dangerous to think of these quick wins as coming for free. Using the software engineering framework of technical debt, we find it is common to incur massive ongoing maintenance costs in real-world ML systems.
A Guide to Production Level Deep Learning - If you're curious how complicated the infrastructure for production level work can get, this is a rundown of a small subset of the ecosystem. It's complex, and a lot of actuaries are misguided to think it's just a click of a button.
Technology for Actuaries in a Digital World - The tone of this article bothers me. I hear all too often people across the financial services and insurance industries treat IT like a liability instead of a differentiator and value creator.
Often there aren’t enough skilled IT resources to support this basic data work, leaving the work to highly qualified actuaries to be the data ‘janitors’.
Perhaps if this is the problem we should fire some actuaries and hire some more IT talent. New tools and technology won't solve what is clearly a cultural problem. His example of taking 120 hours to execute a query screams of an actuary not knowing what the hell they're doing. I'm more concerned that the company needed to hire some consultants to tell them to fix it.
Financial-Models-Numerical-Methods - An awesome resource for anyone interested in combining financial theory with programming in Python.
The Bus Ticket Theory of Genius - I continue to enjoy Paul's writings. (If you haven't read anything by him, you should go through his old stuff.) I asked this question found in the article to multiple people:
If you could take a year off to work on something that probably wouldn't be important but would be really interesting, what would it be?
I didn't give them any context prior, but it was interesting to hear their thoughts. It was especially telling between those who had ideas of problems they wanted to solve and those who didn't. Neither is “right,” but more of a reflection of personality traits that Paul touches on in the article. Highly recommend taking the time to read this one.
The Socialist Revival - The dichotomy between capitalism and socialism has a long history in the American political system. This article gave me some additional avenues to explore.
The Rise and Fall of the Cash Railway - I had no idea something like this ever existed.
Integration and Monopoly - Ben always has good perspective and thoughts regardless of if I agree with them. In this particular case, I think he hits on a lot of the difficulties between defining what is a monopoly.
A History of APL in 50 Functions - I can only pick up bits and pieces of APL, but it's legacy shouldn't be understated.
The Magical Excel 97 Far East Language Build Screwdriver™ - You should take the time to read and enjoy this article.
A Practical Guide to State Machines - State machines are intuitive and useful for all kinds of things. This is a pretty good introduction to them.
Enterprise™ - This had me cracking up for an embarrassing amount of time.
5 Things I’ve Learned in 20 Years of Programming - Any modeling actuary could learn a lot from this.
- Duplication of Knowledge is the Worst.
- Code is a Liability.
- Senior Developers: Trust but Verify.
- TDD Is Legit, and It’s a Game-Changer
- Evidence is King
The care and feeding of software engineers (or, why engineers are grumpy) - A lot of software engineering concepts apply to other technical fields, and this post is no different. Anyone working in actuarial or finance can relate to something in this post. A few quotes that stood out to me:
Nicholas, you’re worth more than your code. Whatever your next gig is, make sure that you’re not a short-order cook. Don’t accept a job where you’re told exactly what to build and how to build it. You need to work somewhere that appreciates your insights into the product as well as your ability to build it.
and
So, without enough information, changing requirements, not enough knowledge to do the job, and people constantly second guessing us, we trudge into work every day. Being creative people, we put up with all of this because we know that one day people will use our work.
A relevant xkcd comic:
That's it for this week. Again, if you have anything you want to share or have any comments, shoot me an email at links@houstonp.com.