May 23, 2024

“What WAS Nimitz Thinking?”: Another Battle of Midway Wargame Analysis

RockyMountainNavy, 5 September 2022

One of the most influential books on the Battle of Midway has to be Jonathan Parshall and Anthony Tully’s Shattered Sword: The Untold Story of the Battle of Midway (Potomac Books, 2005). The book is hailed as a fine example of operations analysis used to cut through historical hyperbole and mythology that grew up around the Battle of Midway. In the Spring 2022 edition of the Naval War College Review, Parshall revisits his famous work in a new light; specifically looking at Admiral Nimitz’s decision to fight at Midway.1 In doing so Parshall engages in a bit of some “what-if” wargame scenario development. Parshall ultimately uses operations research analysis to “wargame” his alternate battles, but in doing so he actually demonstrates the strength of wargames over simple operations analysis. 

In Shattered Sword, Parshall and Tully conclude:

“If one believes in the notion of overwhelming Japanese superiority, then Nimitz’s decision to engage the enemy and accept the horrific odds against him must be judged reckless in the extreme. Nothing less can explain his willingness to walk clear-eyed into a fight, pitting his allegedly pathetic force against the Japanese juggernaut to contest a speck of land that was entirely disposable and that could be isolated and recaptured at any time. However, we take the view that Nimitz was an exceptional commander who had a finer appreciation of the odds facing him than many commentators do sixty years after the battle. Based on estimates of four to five Japanese carriers, he was within his rights to suppose that his forces, if positioned correctly, could carry the day.”2

Parshall goes on to discuss how even his understanding of history has changed in the years since Shattered Sword was published. Specifically, Parshall cites the work of John P. Lundstrom and his book Black Shoe Carrier Admiral: Frank Jack Fletcher at Coral Sea, Midway, and Guadalcanal (Naval Institute Press, 2006). What caught Parshall’s attention is the possibility that the Japanese may have brought a fifth carrier (Zuikaku) In particular, Parshall points out that he learned from reading Black Shoe Carrier Admiral that:

“Not only was Nimitz willing to fight a potential five enemy carriers with three of his own; it turns out he was willing to give battle at odds of five against two, if Yorktown could not be repaired in time from the damage it had suffered at the Battle of the Coral Sea. To my mind, five carriers on three already felt dicey; five on two honestly seemed reckless. And yet this issue has not been addressed squarely in any major history of the battle. What on earth was Nimitz thinking by accepting those odds? And what likely would have been the outcome had such a lopsided battle actually taken place?”3


Paper Time Machines…

Wargames on the Battle of Midway tend to take one of two approaches; either a campaign approach that depicts the lead-up to the battle with an emphasis on maneuvers of the forces and the air search or a more tactical approach that focuses near-exclusively on the actual air battles. Some examples of the campaign wargames taken from my collection are C.V.: A Game of the Battle of Midway (Yaquinto, 1979) or Victory at Midway (XTR Corp., 1992) or Second World War At Sea: Midway (Avalanche Press, LLC, 2002) or The Fires of Midway (Clash of Arms, 2010). On the tactical side, some wargame examples are Kido Butai: Japan’s Carriers at Midway (Dr. Richter Konfliktsimulationen, 2016) or Fury at Midway (Revolution Games, 2020).

IMG 3164
C.V. (Yaquinto, 1979)


In Parshall’s article “What WAS Nimitz Thinking?,” he talks in detail about Nimitz’s battle plan before “gaming” out the alternatives. While many wargamers reading this article might want to focus on the resolution of the battle, I believe that it is far more interesting to look at the decisions that led up to the carrier air strikes. The decisions Nimitz made leading up to the battle and in the early stages before the first air strikes were flown are best replicated in a campaign-level wargame, and not through an operations analysis spreadsheet drill like too many tactical Battle of Midway wargames are. 


Midway Campaign

Parshall makes two key observations about the Battle of Midway that have a tremendous impact on campaign-level games; fleet submarines and Point Luck.

Parshall discusses Nimitz’s plan to use fleet submarines to attrit Kido Butai before the carrier battle. Fleet submarines along with a reinforced air group on Midway was for “strong attrition” intended to whittle down Kido Butai and even the odds going into a carrier engagement.4 

Further, in Nimitz’s plan the U.S. carriers were placed at the famous “Point Luck” located at 32N 173W. From here the U.S. carriers were to monitor the attrition battle and if it went well then commit to battle. Parshall points out that Point Luck was located approximately 360 nautical miles (nm) from where Kido Butai was projected to enter the range of Midway’s aircraft. Given the range of U.S. carrier aircraft, Parshall describes how U.S. carriers would have to steam 8-9 hours from Point Luck to get into a position to launch strikes. This means the earliest the carrier airstrikes could be launched was late-afternoon, not morning. More likely, Parshall assess, Day 1 of the Battle of Midway would have turned into a day used to reposition forces. As Parshall explains, Point Luck was a risk-management tool; it gave the U.S. carriers under Fletcher time and distance to decide if they were going to commit to launching carrier airstrikes. Nimitz was not committed to an all-or-nothing defense of Midway. Specifically, in Nimitz’s orders to Fletcher and Spruance he stated that there was no desire to “slug it out” with Kido Butai if the U.S. position was unfavorable.5 

The combination of submarine attrition and the risk-management of Point Luck barely appear in campaign wargames. Submarines are part of C.V. and SWWAS: Midway but even in those games they are highly abstracted. Fleet submarines make no appearance in Victory at Midway nor The Fires of Midway. Without the fleet submarines, those games rely solely on a reinforced air group at Midway to conduct that “strong attrition.” Maybe not strong enough?

As it turned out, on June 2 Nimitz made a “suggestion” to Fletcher to move the U.S. carriers 175 nm to the west of Point Luck. The impact of this move was to place the U.S. carriers about 185 nm northeast of the projected entry point of Kido Butai. This cuts the whole extra day of steaming into position before a strike that Point Luck demanded. It also was a significantly more risky position. Then again, as Parshall points out, by this time American signals intelligence was strongly indicating that the fifth carrier, Zuikaku, was not part of the invasion force. Maybe this risky decision was not as risky as it looks?6

ship zuikaku15
Zuikaku on her first day of service, 25 Sep 1941 (Courtesy


Alternate Wargame?

In a bit of alternate history, Parshall then explores a “what-if” scenario with five characteristics:

  • the Japanese work more diligently to assemble a composite air group for Zuikaku and commit it to battle at Midway after all, whereupon
  • the first day of battle opens, with U.S. Army Air Forces (USAAF) B-17s making wildly inflated claims of success (which, in fact, they did during the actual battle), thereby
  • convincing the American commanders at Point Luck to commit their carriers to battle on Day 2, only to discover belatedly that
  • Kidō Butai actually has five undamaged carriers, whereupon the Americans would find themselves involved in a carrier action at very unfavorable odds.7

Note that this entire “scenario” is one plausible outcome of a campaign wargame that uses the original Point Luck starting position for the U.S. carriers.


Tactical Midway

Unfortunately for tabletop wargamers, Parshall uses some operations research analysis to play out this what-if scenario instead of a wargame. Inspired by a 2020 article in Military Operations Research, Parshall engaged the authors of the article “Revisiting the Battle of Midway: A Counterfactual Analysis” to use their stochastic model, based on the “salvo combat model” developed by the great CAPT Wayne P. Hughes, to fight this altered Battle of Midway.8 

This is the general situation one faces when playing the tactical wargames like Kido Butai or Fury at Midway. Those two games are in many ways a manual, tabletop version of the “salvo combat model.” But to make try to make it interesting, Parshall explores a tactical wrinkle; that being the performance of the carrier Hornet.

Fury at Midway (Revolution Games, 2020) – Tactical Midway


Good Hornet / Bad Hornet

In the real Battle of Midway, Hornet was commanded by Captain Marc Mitscher who, in a bit of mirror-imaging, sent his airwing on the infamous “Flight to Nowhere” looking for a second Japanese carrier task force. At this point in the war, the Americans did not understand that Kido Butai operated as a combined carrier fleet to better wield its airpower. Mitscher was sure that the Imperial Japanese Navy would split its carriers into separate task forces to strike Midway and provide a covering force. Mitscher believed this covering force would be north of the other carriers. Mitscher’s actions, which in effect removed Hornet’s airwing from a first strike, is what Parshall calls “Bad Hornet.”9

Official U.S. Navy Photograph, now in the collections of the U.S. National Archives.


The Yorktown Factor

Parshall claims the presence of Yorktown decisively tipped the scales in favor of the U.S. carriers. Unfortunately, Parshall does not disclose the data for a 4 vs. 2 battle, but in a 5 vs. 2 battle (i.e. no Yorktown) the Americans come off very poorly; in a “Bad Hornet” situation the Americans will (at best) sink only two Japanese carriers before losing both U.S. flattops. Even if “Good Hornet” is used the Americans will sink two (maybe three) Japanese carriers while again losing both U.S. carriers. I strongly suspect that if one was to play Fury at Midway or Kido Butai without Yorktown the results would be very similar.

Official U.S. Navy Photograph, now in the collections of the U.S. National Archives.


What Are YOU Thinking?

While there is a certain joy in Battle of Midway tactical wargames (after all, who doesn’t like making little dive bomber sounds as you roll the dice?) the fact of the matter is that the “salvo combat model” is a stochastic model and few player decisions have an impact. Parshall himself has stated that the Bongers and Torres model is optimized to replicate the “historical” result. To me, this is more Modeling & Simulation than wargame. Given my druthers, I would much rather play a campaign-level Battle of Midway game before reaching the point of actual battle and then running the model.

A campaign-level game of the Battle of Midway can lead to a very interesting exploration of true what-if situations. Parshall acknowledges that though the stochastic model seemingly indicates an American victory in many cases (outside of the 5 vs. 2 match-up), conditions set by campaign-level games might have a greater impact:

“It is worth pointing out again that in all these scenarios, any improvement in Japanese scouting from the historical norm, which might lead either to a Japanese first strike or even a simultaneous exchange of air strikes, likely would have been disastrous for the Americans. Every carrier in the battle—Japanese or American—theoretically potentially possessed sufficient firepower to disable two enemy flight decks, under the right circumstances. This, in turn, highlights the tremendous importance of good scouting, which, one hopes, confers its most sought-after benefit: allowing one to get in the first effective attack against the enemy, thereby degrading his firepower from the outset. This crucial need to strike first already was well-known at the time of the battle, of course—American prewar exercises had demonstrated this point time and again. This, in turn, explains Spruance’s real sense of urgency and impatience as TF 16’s painfully slow launch cycles were unfolding during the morning of the actual battle. In other words, with either worse American scouting or better Japanese, Midway might have become an American disaster.”10

Indeed, even Parshall concludes that factors outside of the stochastic model were great contributors towards the American victory:

“Chester Nimitz ended up prevailing on 4 June 1942. Where there were problems with his plan, they were offset by even worse Japanese planning and reconnaissance, which ended up wrong-footing Nagumo from the get-go. Nimitz also was aided by the flexible leadership of both Fletcher and Spruance, who were aggressive when called for but prudent at need. These advantages, combined with the skill and bravery of American soldiers, sailors, and airmen—and a very healthy dollop of good luck—were sufficient to achieve victory against a seasoned enemy…Much must be risked in war, and nothing great can come to those who risk nothing. But once the dice are rolled, small changes in circumstance can have very large impacts on the verdict of history.”11

In Battle of Midway wargames, those “small changes in circumstances” are best explored in campaign-level wargames and not tactical-level gaming. Tactical-level games tend to be too close to modeling & simulation and, being also closely tied to preset conditions and based on models optimized towards “historical” results, certainly can provide insight into the battle but ultimately are too deterministic. For a greater understanding, campaign-level games are better suited to exploring those “small changes in circumstances” to discover the “very large impacts on the verdict of history.” Parshall explained to us what Nimitz was thinking, but what do YOU think?

(Feature image courtesy


Thank you for visiting The Armchair Dragoons and saddling up with the Regiment of Strategy Gaming.
You can find our regiment’s social media on Facebook, Twitter, and YouTube.  (We have an Instagram page and we never use it.)  We also have our Patreon, where you can support The Armchair Dragoons activities.
Feel free to talk back to us either in our discussion forum, or in the comments below.



  1. Parshall, Jonathan B. (2022) “What WAS Nimitz Thinking”,” Naval War College Review: Vol. 75: No. 2, Article 8
  2. Parshall, Jonathan B. and Anthony P. Tully, Shattered Sword: The Untold Story of the Battle of Midway (Washington, DC: Potomac Books, 2005), p. 435
  3. Parshall, p. 93
  4. Parshall, p. 99
  5. Parshall, pp. 99-102
  6. Parshall, pp. 106-107
  7. Parshall, p. 108
  8. Bongers, Aneli and Jose L. Torres (2020) “Revisiting the Battle of Midway: A Counterfactual Analysis,” Military Operations Research: Vol. 25: No. 2, pp 49-68
  9. Parshall, p. 110
  10. Parshall, p. 116
  11. Parshall, p. 118

11 thoughts on ““What WAS Nimitz Thinking?”: Another Battle of Midway Wargame Analysis

  1. One options, the USN did not weigh heavily, was to avoid the IJN carriers, and to go after the troop and supply ships. Sink the troop transports, there can be no invasion of Midway in June of 1942.
    Even if the IJN tries again, the USN Saratoga is now available for the battle.

    1. Parshall makes the argument that Nimitz was never concerned about the amphibious landing. Nimitz’s advantage was knowing when and where Kido Butai was going to be and he strived to take advantage of that condition and render it impotent. Nimitz’s plan was always to sink Japanese carriers first.

    2. Even if the Japanese could have overcome the defenses of the island and seized the airfield, that would then mean they would have to support & supply the island, and any American carriers would be in the position of being able to strike (or not) the Japanese supply lines and/or carrier forces at will, while the Japanese would be trying to run convoys all the way from Kwajalein or further.

      1. The invasion force was far enough away from the IJN carriers to try this attack. The invasion force approached from the south-west, and not the north-west, where the IJN carriers appeared.

  2. Please read Midway Inquest by Dallas W. Isom because you will learn at least one critically import decision that is not mentioned in this fine article.

  3. I appreciate the interest in the article; it was a lot of fun for me to take a deep dive into the sources and really try and wrap my head around Nimitz’s mindset.

  4. Having read and dissected the Aneli Bongers & Torres article, I think it’s so circular in argument that it’s at most an reenactment. The randomized part is only the trivial hits or misses at individual plane level. It assumes an US first strike as baseline (without any need for a decision on levels of search effort for each side; and a simultaneous strike as a very remote what-if) and effectiveness percentages derived from Midway alone. [I think the proper Bayesian approach should have pooled data from at least 5 actions, Coral Sea to Eastern Solomons, Or not?]

    Did Nimitz rely surely enough there will be an effective US first strike? i.e. if not, was a simultaneous strike acceptable for the US plan? after all, that happened at Coral Sea.
    Did the whole US staff know they had such superiority in damage control when entering the action? Did they know the Thatch Weave would give them such a fighter tactical superiority ?
    Did the Japanese know any of these? did they make enough effort to improve?

    1. Great points! I think you hit upon reasons why I favor wargaming over Operations Research Analysis; ORSA’s focus on collecting data and then making sense of it whereas wargaming focuses on the decisions and their impacts. The end result from an ORSA is often a “reenactment” that demonstrates the data collected validates the known outcome. If you really want to explore a what-if question, then it’s a wargame you should use.

  5. Just found this excellent post and thread after looking for naval games to get into. As someone who used to do operations research, I’d…wholeheartedly agree with the points made by RMN. OR, or at least the flavor of it discussed here, is akin to a statistical model. Conditional on the data at hand, it produces estimates of the role or influence of different factors. It’s a great tool, but, like any tool, becomes less useful beyond what was designed for. Wargamming (or more academically speaking, counter-factual analysis) can challenge that conditional statement and create a range of what-ifs. Ideally, you’d want to then apply an OR model to those to create a quasi-Monte Carlo analysis. The reason that doesn’t happen much is the effort involved in making the what-ifs and the sometimes difficulty in producing relevant non-historical data for an ensuing OR simulation.

    Having said that and admitting that know very little about Midway, it does seem like there’s enough accumulated research, modeling, and interest to warrant that approach by an enterprising history grad student or another looking into decision-making under uncertainty.

Tell us what you think!

This site uses Akismet to reduce spam. Learn how your comment data is processed.

%d bloggers like this: