Testing Performance, Political Unrest, and New Books
Happy Tuesday, Folks.
It feels a little weird to be publishing a newsletter, given the recent events in the U.S. over the past week. I’ll do something I usually don’t do, which is address a political topic, and then move on with the normal newsletter.
Consequences
”Falsch,” my German teacher corrected me.
I’ve been learning German off-and-on for about 13 years. I’m not that high of a level. I had restarted learning the language a few years ago and was attending classes again.
This was maybe a few years into the Trump presidency, I don’t exactly remember when. The feeling of having a crisp, decisive, “That is not how you do that” nearly knocked me out of my chair. Maybe I was in a weird mood.
For weeks, months, and years at that point, we had been societally been asking “What are words?” and “What is truth, really?” We were witnessing the weaponization of bad faith arguments on a grand scale. Having a boolean answer to something that wasn’t a computer tickled my engineer brain.
It’s difficult to pinpoint what led us to this point. A gradual erosion of critical thinking and trust in the scientific process seem to be contributing factors. The unchecked amplification of political propaganda via social networks doesn’t help either. It led us to a place more violent and more extreme than we were a few years ago.
And we’re watching our society and business respond in different ways. It appears the current outgoing administration has broken the one unwritten rule of American society which is, You do not fuck with market stability or quarterly earnings. The events at the U.S. Capitol were destabilizing events.
I think because these events threaten the very fabric of business in the U.S., companies are pulling the trigger on things they’ve been waiting to do for some time: They are deplatforming political individuals that haven’t had support of the employees within the company. This is an extraordinary measure to some degree, but it is justified in my mind. These are extraordinary times, and the only sin would be to continue with business as usual.
Last week I subtweeted what I was watching: “Are we witnessing chapter 2 or chapter 10? Is this the end of the beginning, or the beginning of the end?”
I think what we’re seeing is a political party in decline. In a society based on the stability and predictability of the markets, a political force that is becoming more violent, more extreme, and fewer in numbers, is not long for this world. Demographically, they’ve been losing for some time. Ideologically, they have become so inconsistent it’s dizzying (e.g. “Law & Order” and “Beating a police officer to death with the American flag is peaceful protest.”)
As we collectively disengage from our System 1 brains and move back into System 2, we’ll probably see a rebirth of the GOP. If I had to guess, it will come from some of the more hypercapitalist ideologies. Indeed, the hypercapitalist layout of American society seems to be the only thing holding the president in check at the moment (i.e. deplatforming him and others). The checks and balances within the government have failed.
Every two-party country needs principled majority and minority governments. We need a party to be a critical voice to point out the ineffectiveness of just vaccine policies, like those seen in New York. (Octogenarian individuals have to fill out a long questionnaire, upload documents, etc. to qualify themselves for a vaccine.) Right now, the soon-to-be-minority government is busy convincing its followers that democracy shouldn’t exist and that COVID-19 is a lie.
Nonetheless, I hold out hope that we’ll correct this in the next decade. The most violent and most extreme will eventually fade. Looking to distance itself from anything related to the years 2016-2020, the GOP will have to create a new identity. And if I had to further guess, this might result in places like San Francisco flipping red sometime in my life.
Testing Performance
This week’s blog post is on testing performance in applications.
Over my career, I’ve always had trouble writing tests that captured regressions such as, “This change will make this endpoint 3% slower.” These types of benchmarks are noisy and often lie to us.
Furthermore we don’t know if that 3% will crash the production system or if it won’t even be realized at all because the bottleneck in test is not the same as production.
My recommendation is instead to write tests that qualitatively test code as a means of predicting performance changes. Questions like, “Does this query use an index?”
Read the post and let me know what you think.
New Books
I’ve reorganized my site to list almost all of the books on my bookshelf at the moment, along with a few other miscellaneous resources. This is an expansion of the “Reading List” I had previously had.
This was inspired by Patrick Collison’s bookshelf on his site. I’ve picked up a few from that list and have been enjoying them so far.
This week, I cracked open Capital and Ideology. I’ll likely do a writeup on my blog or in a future newsletter.
Thanks for reading, and stay safe out there.