# Systems research requires a lot of work that is expensive and uninteresting When you’re trying to change a large system, a big chunk of the work doesn’t involve the actual interesting bits that you’re changing. Instead, you need to spend a lot of effort propagating changes through the entire system. From a software framing, it’s like making a change that requires you to touch a majority of the files in a large repo in a mindful way. Switching a car from an internal combustion engine to an electric motor requires not just swapping out the obvious things (like a battery instead of a gas tank) but rearchitecting the entire car — what do you do with the space where the engine used to be? How do you deal with the very different center of mass? How do you do steering when motors are at the wheels? How do you do refueling? Figuring out all of these answers requires work.[^1] From [[pikeSystemsSoftwareResearch2000]]: `Estimate that 90-95% of the work in Plan 9 was directly or indirectly to honor externally imposed standards.` Academia is especially allergic to work that simply propagates changes throughout a system because [[Academia incentivizes novelty, not focus]]. While the initial change might check a novelty box, propagated systems changes are not in themselves novel, driving down the publishable work/total work ratio. The uninteresting work involved in systems research is one reason why [[A lot of research feels like it is building lego pieces without any idea how they’ll be used]]. You see lots of “component demos” that show how a tweak to one piece of a system or another could potentially have an effect. Theoretically an organization that does systems research could pick up that change and propagate it through a system, but that rarely happens. Other institutions have their own incentives against systems research. It’s painful but straightforward to eat the pain if you know beforehand that a systems-level change is worthwhile. However, the nature of research means that it’s not clear from the outset which changes are going to lead to a better system and even if you do know which ones will lead to a better system in the long run, [[Systems must take performance hits to get out of local optima]]. As a result, a lot of the expensive uninteresting work doesn’t directly improve the system and may in fact be completely “wasted” or unjustifiable. This waste runs headlong into cultures that prioritize justification and efficiency. The inherent inefficiency of systems research is another reason why [[Efficiency is overrated]]. Propagating changes through a system is why [[Systems Research]] takes a surprising amount of time. Point changes to a system are O(1) (or maybe O(N)) while systems level changes are somewhere around O(N^2) because in the worst case scenario you need to pay attention to every pairwise interaction between every two components. The expensive, boring work in systems research is more often justifiable when there is a clear metric for the system -- another way that [[SpaceX]] and [[Bell Labs]] were powerful. However, one key point of disruption theory (see [[christensenInnovatorDilemmaWhen1997]]) is that new systems often win on completely different metrics than the ones they replace. This metric shifting is also core to paradigm shifts in science. ([[Metrics can cause paradigm lock-in]]). In the absence of a metric, a [[Serious Context of Use]] is important for [[Systems Research]] in large part because it makes sure the expensive uninteresting work happens. A context that someone cares about casts parts of the system that need to be updated based on changes elsewhere into a harsh light — “you made a change over here that broke this other thing I really care about.” Without a context that someone cares about, it’s easy to ignore how the change of interest affects other parts of the system. ### Related * [[Governments rarely fund systems research unless it leads directly to a widget]] * [[Efficiency biases systems towards false negatives]] * [[Startups are good at point changes]] * [[We don’t have good institutional ways to create multi-point changes]] * [[What does ‘working’ mean?]] * ### References * [[fillerFundamentalManufacturingProcess2020]] * [[raoGUTSGrandUnified2018]] [^1]: The systemic changes needed by hydrogen cars may be one reason why electric cars appear to be winning — at least we already have the infrastructure for electricity distribution. Of course there are lots of other factors.