The orange man is bad. Yes, that’s really simple. It’s caveman logic. But it outperforms the more sophisticated takes.
The only consistently valid model of Donald Trump is that he is a malevolent actor obsessed with grabbing power and buffering his ego. He is indifferent to the health of the system and quite frankly also indifferent to the need to accept a world that exists outside his own degenerating self-consciousness. His presidency is a bubble and there is going to be chaos when that bubble bursts. You get the picture.
Orange man bad outperforms more sophisticated models, but it also doesn’t get you far on its own. There are a number of analytical pitfalls that are easy to fall into, even if they aren’t entirely insoluble. This post talks about the partial observability of administration decision-making and its consequences for interpreting executive decisions.
On partial observability and observational equivalence
The reason why spy services place an extremely high value on secret intelligence in a world of open source data is that the elite processes that produce political decisions always occur out of the public eye. If those processes were transparent, if whether Premier is bluffing could be predicted with certainty, wars likely would not occur at all. Clandestine intelligence is a premium good because Kremlinology precise enough for high-stakes scenarios is far from trivial.
When you don’t know much about the actual structure generating the decision, Tom Pepinsky pointed out, it gets even harder. We can only observe the output of Trumpian behavior. We don’t observe everything that goes into those decisions and have little idea how they are generated. Therefore the same “data” can often support multiple contradictory theories. Trump administrative decision-making has always been opaque. One always risks falling prey to a kind of “efficient Trump theory” in interpreting it, but underrating the administration is also unwise as well.
These kinds of issues might not be much of a problem if we could at least get enough post-hoc understanding of decision structure to evaluate models through retrodiction. Pepinsky, for example, mentions biographies of Eisenhower and Bush II that critically reinterpret history in light of new evidence and approaches. But that’s the rub. What makes — in both Trump I and Trump II — decision-making so mysterious is the extreme instability and mutability of the decision-making structure itself. The Trump administration seems tailor-made to frustrate political analysts.
I can’t really improve on Venkatesh Rao’s comment four years ago about the fine line between attributing causal structure to Trumpian behavior and justifying that behavior:
Post hoc arguments justifying market outcomes are eerily similar in structure to arguments justifying Trumpian outcomes. Like they believe in some sort of efficient Trump hypothesis… Looking for causal models here, let alone exclusively determinative one, is as misguided. The market and Trump do not work on mechanistic causation but via emergence from information pricing… hurricanes are not “caused” by individual gusts of wind lining up but heat and now pressure regions. In a way, Trump is the notional “rogue field commander” envisioned in brinkmanship scenarios. He just happens to be at the top.
The ambiguity of who is in charge and how decisions are made still makes key moments of Trumpian history mysterious, going well beyond analytical problems of other personalist regimes. We do not know, for example, a great deal about the June 2020 and January 2021 periods, despite their tremendous importance to the past , present, and future issues we are thinking about. This is, in some ways, one of the biggest absurdities of the Trump presidency. Something so embarrassingly public and leak-prone is still very much occluded. It is erratic but more importantly non-stationary.
Essence of indecision
Let’s start with the President alone. I wrote during the first Trump administration that Trump channels the mutually contradictory feelings of the people who follow him and the hatred of his enemies, reacting to the chaotic stimuli around him while simultaneously driving the things influencing him to new levels of derangement and instability. There are certainly historical analogues (“working towards the Fuhrer”) but the Trump I and II terms take dynamics within them to surreal extremes. Some of them are described here, others will have to wait until later.
As an individual, the infamously mercurial and impulsive Trump might be plausibly considered as multiple agents inside a single body instead of a single person. Arbitrary environmental cues have significant influence — Trump watches a Fox News report and then suddenly issues a 2AM directive, Trump’s papers are snatched off his desk, Trump gets flattered by a foreign leader, and so on. Those cues change over time. Trump is also increasingly senile, so the biological internal setup is also growing more disorderly. When detailed medical data is finally available for historical use, our understanding may change with it.
Trump is also a weak President. He has never been interested in accepting the duties of his office, using its powers, standardizing a decision process, and most importantly setting a clear chain of command. This means that there is no inherent bound on the qualitative and quantitative variety of actors with inputs into an executive behavior. Anyone from a random social media meme maker to the Secretary of Defense has a theoretically equal chance of influencing a decision or its execution. With no stable configuration and strongly subject to context, the decision structure always mutates when it decides.
Elon Musk’s role — and that of DOGE more generally — has also introduced another strange element to Trump II. Trump is increasingly becoming a nonentity, Musk looks like a shadow President, and there’s a teenage cybercriminal nicknamed “Big Balls” running loose in the most sensitive government computer networks. It’s too early for us to have a well-developed understanding of Musk and DOGE and most information about them is partially compromised. Nor is Musk’s role — like anything else Trump-related — guaranteed to remain stable over time.
A final novel chaotic factor introduced by the second Trump term is that there is now a seamless fusion between “normal” political and geopolitical events with the following processes:
- Media news cycles (frequently pseudo-events)
- Social media attention bubbles
- AI-generated meme crazes
- Dropship consumer fads
- Criminal scams (MLM and more)
- Financial market movements
- Cryptocurrency markets
- Prediction markets
- Pump and dump schemes
So there are actors orbiting the administration (and likely within it) that have a financial motive for triggering or at least profiting from short-term political crises. The heavily automated character of all of this — whether speaking of high-frequency trading on Wall Street or the progressively increasing colonization of online spaces by user-emulating computer programs — also greases the wheels of events that already occur at blindingly fast speeds. A grotesque inversion of the Keynesian Beauty Contest is going on during every hour, minute, second, and micro-second. 1
You, too, dear reader, are also likely a minor contributor to the idiotic maelstrom. If you follow each event of the Trump era obsessively in the news and social media, you see a recurring process play out over and over again. There’s a moment of uncertainty and all explanations, predictions, and scenarios are entertained simultaneously. For a brief shining moment, everything is truly possible. Then the waveform collapses and a single outcome emerges. All other possibilities are instantly forgotten. Naturally, we knew it all along! Never mind the crazy theories we all just posted!
Yes, there’s more
Readers may be tempted to scoff at the utility of all of this over-analysis, but please defer those objections until later. The next several posts in this series are going to talk more directly about the contributing role of public commentary and standards for evaluation of theories and models.
Footnotes
-
If you are looking to understand how this process works from fictional standpoint, consult Phillip K. Dick novels, David Cronenberg films, the animes Serial Experiments Lain and Ghost in the Shell: Stand Alone Complex, and the game franchises Metal Gear Solid 1-4 (with emphasis on MGS2). as well as Killer7. ↩