Enemy Lines By Bill Doyle

Enemy Lines By Bill Doyle: A Practical Guide

You pick up Enemy Lines By Bill Doyle because someone in your office said you needed to understand adversarial thinking better. That usually means your project got complicated fast and the other side isn't cooperating with your assumptions. The book walks through a methodology for mapping out how opponents think, what they value, and where their decision chains break. It isn't theory for theory's sake. Doyle builds it from cases he actually worked through during his intelligence analysis career. The method centers on building an adversary model through structured analysis rather than gut feeling. You identify the decision-makers, map their incentives, and then trace how information flows through their organization. The key move is separating what the enemy knows from what they want. Most people conflate those two and end up with models that look accurate on paper but fall apart when reality shows up. I spent weeks once trying to forecast how a specific regulatory body would respond to a product launch. I had the public statements, the press releases, the political commentary. Everything looked consistent until I realized I never actually mapped who reported to whom and what metrics those individuals were evaluated on. Once I did that — which took about three days of digging through organizational charts and earnings calls — the predicted response shifted entirely. The person with veto power wasn't the loud one giving interviews. It was a mid-level operations head who answered to cost targets, not public optics. Enemy Lines By Bill Doyle gives you the template for exactly this kind of structural breakdown.

How to Actually Use the Method

Start by listing every actor you think matters in the adversary's decision chain. Be aggressively inclusive at first. Then rank them by actual influence, not visibility. The people giving press conferences are rarely the ones making the final call. Look for budget authority, approval hierarchies, and performance incentives. Those reveal real power. From there, build what Doyle calls a preference tree. This maps each actor's likely choices across different scenarios. The trick is forcing yourself to write out the worst-case and best-case reasoning for each option, not just the most obvious one. Human analysts skip this step constantly because it feels uncomfortable to articulate why an opponent might do something that seems irrational. They always have a reason that makes sense inside their own framework. Then stress-test the model. Pick three scenarios where your prediction would be completely wrong and write out what evidence would prove you wrong. This is the step most people abandon because it feels like undermining your own work. It's actually the opposite. If you can't articulate what would falsify your model, you don't actually have a model. You have a story you like.

Where the Method Breaks Down

The approach assumes the adversary operates with some internal logic and consistent preferences. That's a reasonable assumption most of the time. It fails hard when dealing with actors who genuinely don't care about self-preservation, institutional stability, or any predictable outcome. I encountered this when modeling a group that seemed to reward public failure over private success. Their incentive structure was inverted compared to everything in the book, and I wasted two weeks trying to force their behavior into a standard framework before I admitted the model didn't apply and switched to tracking observable actions instead of inferred motivations. Another limitation is the time investment. A proper adversary model using this method takes roughly four to six hours for a moderately complex situation. You can rush it down to about ninety minutes, but the quality degrades sharply below that threshold. If your decision timeline is shorter than that and you can't get buy-in to slow down, the method becomes a liability. In those cases, use the rapid assessment variant Doyle mentions in the later chapters — it sacrifices granularity for speed and is more defensible when you're working under pressure.

Practical Tips From Real Use

When building preference trees, always include a column for what each actor stands to lose, not just what they stand to gain. Loss aversion dominates decision-making more than people in most organizations want to admit. I've seen models that predicted action because the potential upside was large, then completely miss because the person involved had a personal consequence they'd avoid at almost any cost. Also, verify your source chain. A lot of analyst work falls apart because the underlying organizational data came from a single blog post or an outdated LinkedIn profile. Cross-reference hierarchy information against at least three independent sources before building it into your model. This usually adds a couple hours to the process but saves you from building your entire analysis on a false premise about who answers to whom. The downloadable worksheet templates Doyle provides are worth using even if you don't follow his exact format. The structured approach forces you to make implicit assumptions explicit, which is where most errors enter the system. I print out the preference tree template and fill it by hand during initial analysis. Writing things down physically slows me down enough that I catch inconsistencies I'd normally skip over on a screen.

One more thing nobody tells you about this method: it works best when you share your preliminary model with someone who has direct experience in the adversary's environment, even if they disagree with your conclusion. Disagreement surfaces weak points faster than any amount of solo review. I usually send my early drafts to one or two colleagues with relevant background and ask specifically what I'm missing, not what's wrong. The framing matters. People are more honest when they don't feel like they're grading your work.