Proxies, Fads, and the Swiss Army Knife Problem in Education
Education has a habit of mistaking shadows for objects. We cannot physically analyze receptors in students’ brains every time they comprehend a metaphor, internalize a math concept, or revise a sentence. So we sample. We pull a subset of items from a larger bank and we try to generate a score. That score becomes a proxy for learning.
That move to proxies is entirely rational, because systems need signals, but proxies tend to have a strange gravitational pull. Over time, they stop being signals about learning and start becoming de facto learning itself. We design curriculum around them, evaluate teachers through them, and rank schools by them. We fund or defund programs because of them. And then we act surprised when the system bends toward the metric.
The story of the SAT is instructive here. The test’s intellectual ancestry traces back to Army Alpha tests developed during World War I to sort military recruits efficiently. It later evolved into the SAT and was positioned as a predictor of college success. That is a remarkable career shift—from sorting soldiers to forecasting academic futures.
It correlates modestly with early college performance, but it reflects cultural familiarity as much as, or more than, raw academic readiness. That’s different from being a comprehensive measure of readiness, talent, or potential. A tool designed for one function can stretch into another, but the stretch is never neutral. Meaning accumulates and assumptions harden. The proxy becomes destiny. Any proxy that correlates with both ability and advantage must be interpreted cautiously. Otherwise, the system quietly rebrands structural familiarity as innate readiness.
This dynamic shows up repeatedly. Value-Added Models had their era. For a stretch of time, VAM was presented as the future, offering precision analytics that could isolate a teacher’s contribution to student growth. Conference stages lit up. White papers multiplied. Then reality intruded, because the models were noisy and context mattered more than we wanted to admit. The miracle did not arrive.
So the system pivoted, and dashboards became the thing. Then when those didn’t lead to miracles, personalized learning platforms were all the rage. And now we are pivoting once again to AI. None of these tools are inherently misguided, and on their own each represents a certain type of innovation. The problem is monoculture. We jump wholesale from movement to movement, treating each new approach as the master key.
Imagine if music worked that way. Five years of jazz combos, then everyone discards it and becomes an expert in ska. Then suddenly everyone insists only electronic music counts. It would be chaos. Music works because genres coexist. Musicians build fluency across styles.
Education, by contrast, tends to grab one tool and insist it is the whole toolbox. This is where proxies and fads intersect. When a proxy becomes dominant, it creates incentives. When a fad becomes dominant, it crowds out alternatives. In both cases, complexity gets flattened.
Now let’s layer in school choice. Choice systems rely heavily on proxies, like star ratings, test scores, growth percentiles, and college acceptance rates. Families who are navigating options need signals, and systems provide them. Again, this is completely rational, but from a cultural capital lens, this becomes more interesting. Cultural capital refers to the knowledge, habits, and signals that help people navigate institutions effectively. Some families (typically middle- and upper-class) know how to interpret data points cautiously. They visit campuses and talk to parents. They read between the lines. Others who don’t possess this understanding may rely primarily on visible metrics because those are the only signals provided.
If the system elevates one proxy above all others, it shapes behavior. Schools optimize for the metric, and families chase it. The metric becomes shorthand for quality. Yet cultural capital is not reducible to a test score. A school’s ability to cultivate intellectual curiosity, social fluency, disciplined thinking, or creative risk-taking rarely fits neatly inside a dashboard tile. When choice operates in a proxy-dominated environment, it subtly trains everyone to value what is measurable over what is meaningful.
I want to be clear: this is not an argument against measurement, but it is an argument for measurement humility. Large-scale standardized testing provides comparability, which matters. But what if we complemented that approach with something structurally different?
Imagine an accreditation-style system with randomized, in-depth inspections. Not compliance theater, but real inquiry. Expert teams who are trained to examine instructional quality, curriculum coherence, student work, school culture, and leadership. Randomized visits reduce performative prep. Multiple data sources reduce overreliance on a single score. Instead of asking, “What is this school’s number?” we might ask, “What is happening here?”
Critics will immediately raise scalability and consistency concerns, which is fair, but those are engineering problems, not philosophical refutations. We already spend substantial public resources on testing infrastructure and analytic contracts. Reallocating even a portion of those funds could support a professional inspection corps.
The objective is to diversify signals and not replace one monoculture with another. Systems that rely on multiple measures are harder to game and more resilient to failure. When one proxy weakens, others still function. When one tool underperforms, others compensate. This is the Swiss Army knife principle. Swiss Army knives are not elegant in the way a single, polished blade is elegant. They are pragmatic and anticipate that no single tool will solve every problem.
Education, however, often behaves more like baking soda, where something is stretched far beyond its original purpose: a student test score becomes a proxy for teacher effectiveness or a ranking becomes shorthand for school quality. The problem is our impulse to repurpose without reflection. Learning is multi-dimensional: cognitive, social, emotional, cultural, and developmental all at once. Teaching is adaptive. Leadership is contextual. Families are navigating layered realities. A pluralistic system acknowledges this and demands intellectual discipline. When a new idea emerges, be it AI tutoring systems, adaptive diagnostics, or competency-based transcripts, the focus shifts to integration. Where does this tool fit, what does it measure well, what does it miss, and how does it interact with existing structures? This is how we find the right tool for the job, rather than forcing existing tools into new purposes.
Proxies will always exist. They are necessary in large systems. And fads will emerge, because innovation is messy and uneven. Education does not need another singular revolution. It needs better calibration. Multiple proxies (interpreted cautiously) for multiple approaches (deployed strategically) producing multiple signals (weighted intelligently).
That may not headline a conference or trend on social media, but systems built on plural tools rather than singular answers are more stable, adaptive, and honest about the complexity they are trying to govern. In a field as complex as education, honesty is a competitive advantage in the long run.
