Two Kinds of Shadow Power

Part of The Build. Start with the guide for the full map.

The decision-rights piece elsewhere in this series names the general failure mode. Influence exercised outside any formally assigned role. This piece is the pattern-level build underneath that framework because I’ve come to think the general finding hides a more useful distinction. It isn’t one failure pattern. It’s two. They fail through different mechanisms and a campaign manager evaluating a role needs to know which one they’re walking into before day one.

Across the prospective 2028 field, is unformalized influence a single, uniform risk, informal power that should just be minimized or are there structurally distinct varieties that call for different responses?

Nearly every campaign this research examines shows some version of the same structural feature, a person with unformalized influence who doesn’t appear on any org chart. A spouse. A sibling. A political operative who’s been in the room since before a campaign was a real possibility. That near-universality holds across the campaigns this project has examined in depth. Within that bounded set, though, the pattern is consistent enough to suggest this isn’t really about any individual candidate’s judgment or insecurity. It’s a structural feature of how modern political careers get built. Careers accumulate trusted people over years, long before a campaign exists to formalize their role in. A campaign launch doesn’t erase those relationships, but instead adds a formal structure on top of them which the informal relationships then have to somehow coexist with.

The first type is a personal relationship with no formal campaign role at all, someone whose influence comes purely from proximity and trust. The mechanism of failure here is ambiguity rather than conflict. Nobody on the formal staff quite knows how much weight this person’s opinion carries which means decisions that should be settled at the senior-staff level get quietly re-litigated in a conversation nobody else is in the room for. This type isn’t usually adversarial to the formal structure. It coexists with it, uneasily, rather than actively undermining it. The failure mode is inefficiency and unpredictability. The same decision might get made differently depending on which informal conversation happened to occur first.

The second type is structurally different and I think it’s the one that gets underweighted in how campaigns typically assess this risk. It’s a longtime political operative embedded in the formal structure, often in a titled role but whose loyalty and communication channel run to the candidate personally and bypasses whatever chain of command exists on paper. This is a parallel authority structure that can move faster than the formal one precisely because it doesn’t have to clear the formal one’s process. From outside the innermost circle, it can look for a long stretch like the formal structure is working normally right up until a contested moment reveals that the real chain of command was never the one on the chart.

I don’t have a clean, named historical instance of this specific type to point to the way type one has one; this dataset doesn’t currently contain a documented case detailed enough to walk through the way the Harris case can be walked through for type one. What follows is argued from structure and mechanism rather than a worked historical example. That’s a different evidentiary footing.

Treating both as “informal influence, minimize it” misses that they call for different responses. Type one is primarily an information problem. The fix is transparency. Give the person a formally scoped role (even a narrow, honest one, like “consulted on family and scheduling matters”) so the rest of the staff has an accurate picture of where real influence actually sits rather than guessing. Type two is primarily an authority problem. Transparency alone doesn’t fix it because the issue isn’t that people don’t know the channel exists so much as the channel structurally outcompetes the formal one whenever the two conflict. The fix there has to go further. Either elevate that person’s formal role to match their real authority so the chart and the reality converge or have a serious, uncomfortable conversation about whether that person’s role should scale down to match what’s actually on paper. Neither type is disqualifying on its own. Plenty of successful campaigns have run with a candidate’s spouse deeply involved or a longtime loyalist in a senior seat. This project has no basis for treating any specific 2028 candidate’s version of this pattern as confirmed rather than a general structural risk that shows up across the field.

An alternative reading is that maybe there’s just one underlying phenomenon, unformalized trust, and what looks like “type two” is just “type one” with a longer history and a formal title attached. I don’t think this fully holds for one specific reason. Type one’s failure mode (ambiguity, inconsistent decisions) and type two’s failure mode (a parallel chain of command that can override the formal one entirely) aren’t different points on the same scale. A highly trusted type-one figure with no title still can’t formally override a titled decision-maker; they can only create confusion about what that decision-maker should do. A type-two figure, by virtue of holding a real formal role, can. That’s a difference in kind, not just degree, even if the two types often originate from similarly long personal relationships.

Not every embedded, personally-trusted operative produces the type-two failure pattern. Plenty function exactly as their formal title suggests with exercised loyalty to the formal process itself rather than a parallel channel around it. The distinguishing factor is whether that person’s own conduct treats the formal structure as binding on themselves. A trusted, embedded operative who routes their own recommendations through the same process everyone else uses is type two in title only; the risk only materializes when that person’s own behavior signals the formal process doesn’t apply to them.

The near-universal presence of some unformalized-influence figure across the campaigns this research has examined is established for that bounded set; a broader claim about the entire historical field would overreach what this dataset supports. The two-type distinction and its differing failure mechanisms is supported-but-provisional. Type one rests on a single, well-documented case; type two is argued from mechanism because this dataset doesn’t yet contain a clean, named instance of it either.

If there’s a single piece of advice in this whole body of research to take in before accepting a senior 2028 campaign role, it’s to find out, specifically, which type you’re walking into. Ask not just “who has real influence here” but “does that person’s own behavior treat the formal chain of command as binding on them.” Being offered a job by someone who turns out not to be the real center of authority is to deal with type one or type two depending on that answer. The answer will tell you more about whether the job is winnable than almost anything in the campaign’s official plan.