Category Archives: Wargames

The Problem With Hexagons

38 Replies

Hexagons are ubiquitous in wargames now (indeed, both Philip Sabin’s War: Studying Conflict Through Simulation Games and Peter Perla’s The Art of Wargaming feature hexagons on their book covers), but this wasn’t always the case. My first wargame – the first board wargame for many of us – was Avalon Hill’s original Gettysburg  (by the way, $75 seems to be the going price for a copy on eBay these days).

No hexagons in Avalon Hill’s original Gettysburg. Remember how the map contained the original starting positions for the Union cavalry and out posts? From author’s collection. (Click to enlarge)

The American Kriegsspiel by Captain Livermore (circa 1882) only had a map grid for estimating distances. We also have a map grid in General Staff to facilitate estimating distances but you can turn the map grid on or off.

Plate 1 from The American Kriegsspiel by Captain Livermore. Click to enlarge. This image is from GrogHeads wonderful blog post on Nineteenth Century Military War Games. Link: http://grogheads.com/featured-posts/5321

And how about this picture from the Naval War College (circa 1940s)? I just needed an excuse to post this photograph:

A Fletcher Pratt Naval War Game in progress. I never understood why they didn’t use upside down periscopes to check broadside angles rather than getting down on the floor. Click to enlarge. From this blog http://wargamingmiscellany.blogspot.com/2016/02/simulating-gunfire-in-naval-wargames.html

It is pretty common knowledge among the wargaming community that Avalon Hill’s owner, Charles Roberts, introduced hexagons to commercial wargaming in the early 1950s .

“Later, he [Roberts] saw a photograph of one of the RAND gaming facilities and noted they were using an hexagonal grid. This grid allowed movement between adjacent hexagons (or hexes, as they are more frequently called) to be equidistant, whereas movement along the diagonals in a square grid covered more distance than movement across the sides of the squares. Roberts immediately saw the usefulness of this technique and adopted to his subsequent games.”

– The Art of Wargaming, Perla, p. 116

In researching how the RAND Corporation – a major post-war defense think tank – came up with the original idea of employing hexagons to simplify movement calculations (as well as the invention of the Combat Resolution Table or CRT) I stumbled upon an amazing document: Some War Games by John Nash and R. M. Thrall (Project RAND, 10 September 1952; available as a free download here). Yes, that is THE John Nash; A Beautiful Mind John Nash; the Nobel Prize recipient John Nash. The Some War Games summary states:

“These games are descendants of the one originally instigated by A. Mood, and are both played on his hexagonal-honey comb-pattern board. – Some War Games Nash & Thrall.

But what appears on Page 1A of Some War Games is even more exciting:

The earliest reference of using hexagons for wargames. “The board is a honeycomb pattern of hexagonal “squares,” the same that was used in Mood’s game” – From Some War Games (Project RAND, Nash & Thrall).

Sadly, I have been unable to find an actual copy or documentation for “Mood’s game,” but did discover that A. Mood was a statistician who wrote the popular text book, “Introduction to the Theory of Statistics,” and, during World War II was involved with the  Applied Mathematics Panel and the Statistical Research Group. Mood was also the author of, “War Gaming as a Technique of Analysis,” September 3, 1954 which is available as a free download here. Unfortunately, I have yet to uncover any images of Mood’s original war game and the very first use of his ‘honeycomb pattern’ board.

Let’s take a quick look at the math behind hexagons:

The cost of moving diagonally as opposed to horizontally or vertically on a map board (from a slide in my PhD Qualifying Exam on least weighted path algorithms).

The problem of quick and easy movement calculation (as shown in the above graphic) is caused by the Pythagorean Theorem. Well, not so much caused, as a result of the theorem:

The distance to a diagonal square, d, is the square root of the square of the hypotenuse (the side opposite the right angle) which is equal to the sum of the squares of the other two sides. We all learned this watching the Scarecrow in the Wizard of Oz, right?

In other words, if everybody could just multiply by 1.41421356 in their heads we wouldn’t even need hexagons! The downside, of course, is now we’ve restricted our original eight axes of movement to six. And there’s another problem; what I call the, “drunken hexagon walk.”

An example of “drunken hexagon walk” syndrome. All we’re trying to do is go in a straight line from Point A to Point B and from Point A to Point C.

In the above diagram we just want to travel in a straight line from Point A to Point B. It’s a thirty degree angle. What could be simpler? How about traveling from Point A to Point C? It’s a straight 90 degree angle. It’s one of the cardinal degrees! What could be simpler than that? Instead our units are twisting and turning first left, then right, then left like a drunk stumbling from one light post to another light post across the street. In theory the units are actually traversing considerably more terrain than they would if they could simply travel in a straight line. This is the downfall of the hex: sometimes it simplifies movement; but just as often it creates absurd movement paths that no actual military unit would ever take.

So, what’s the solution? Clearly, there is no reason why a computer wargame should employ hexes. Computers are very good at multiplying by 1.41421356  or any other number for that matter. Below is a screen shot of General Staff:

Screen shot of General Staff (2nd Saratoga) based on the map of Lt. Wilkinson, “showing the positions of His Excellency General Burgoyne’s Army at Saratoga published in London 1780)  . Click to enlarge.

What’s missing from the General Staff screen shot, above? Well, hexes, obviously. Units move wherever you tell them to in straight lines or following roads precisely if so ordered. And units can obviously face in 360 degrees. Consider this screen shot from the General Staff Sandbox where we’re testing our combat calculations:

Screen shot from the General Staff Sandbox. Notes: 3D unit visibility is turned on, displayed values: unit facing, distance, target bearing, enfilade values, target offset. Click to enlarge.

For board wargames hexagons seem to be a necessary evil unless you want to break out the rulers (that never stopped us with the original Gettysburg or Jutland). But, when it comes to computer wargames, I just don’t see the upside for hexagons but I do see a lot of downside. And that’s why General Staff doesn’t use hexes.

Computational Military Reasoning Part 4: Learning

1 Reply

In my previous three posts on computational military reasoning (tactical artificial intelligence) we introduced my algorithms for detecting the absence or presence of anchored and unanchored flanks, interior lines and restricted avenues of attack (approach) and retreat. In this post I present my doctoral research1)TIGER: An Unsupervised Machine Learning Tactical Inference Generator which can be downloaded here which utilizes these algorithms, and others, in the construction of an unsupervised machine learning program that is able to classify the current tactical situation (battlefield) in the context of previously observed battles. In other words, it learns and it remembers.

‘Machine learning’ is the computer science term for learning software (in computer science ‘machine’ often means ‘software’ or ‘program’ ever since the ‘Turning Machine'2)https://en.wikipedia.org/wiki/Turing_machine which was not a physical machine but an abstract thought experiment.

There are two forms of machine learning: supervised and unsupervised machine learning. Supervised machine learning requires a human to ‘teach’ the software. An example of supervised machine learning is the Netflix recommendation system. Every time you watch a show on Netflix you are teaching their software what you like. Well, theoretically. Netflix recommendations are often laughingly terrible (no, I do not want to see the new Bratz kids movie regardless of how many times you keep recommending it to me).

Unsupervised machine learning is a completely different animal. Without human intervention an unsupervised machine learning program tries to make sense of a series of ‘objects’ that are presented to it. For the TIGER / MATE program, these objects are battles and the program classifies them into similar clusters. In other words, every time TIGER / MATE ‘sees’ a new tactical situation it asks itself if this is something similar (and how similar) to what it’s seen before or is it something entirely new?

I use convenient terms like ‘a computer tries to make sense of’ or a ‘computer sees’ or a ‘computer thinks’ but I’m not trying to make the argument that computers are sentient or that they see or think. These are just linguistic crutches that I employ to make it easier to write about these topics.

So, a snapshot of a battle (the terrain, elevation and unit positions at a specific time) is an ‘object’ and this object is described by a number of ‘attributes’. In the case of TIGER / MATE, the attributes that describe a battle object are:

  • Interior Line Value
  • Anchored / Unanchored Flank Value
  • REDFOR (Red Forces) Choke Points Value
  • BLUEFOR (Blue Forces) Choke Points Value
  • Weighted Force Ratio
  • Attack Slope

The algorithms for calculating the metrics for the first four attributes were discussed in the three previous blog posts cited above. The algorithms for calculating the Weighted Force Ratio and Attack Slope metrics are straightforward: Weighted Force Ratio is the strength of Red over the strength of Blue weighted by unit type and the Attack Slope is just that: the slope (uphill or downhill) that the attacker is charging over.

TIGER / MATE constructs a hierarchical tree of battlefield snapshots. This tree represents the relationship and similarity of different battlefield snapshots. For example, two battlefield situations that are very similar will appear in the same node, while two battlefield situations that are very different will appear in disparate nodes. This will be easier to follow with a number of screen shots. Unfortunately, we first have to introduce the Category Utility Function.

So, first let me apologize for all the math. It isn’t necessary for you to understand how the TIGER / MATE unsupervised machine learning process works, but if I don’t show it I’m guilty of this:

The Category Utility Function (or CU, for short) is the equation that determines how similar or dissimilar too objects (battlefields) are. This it the CU function:

‘Acuity’ is the concept of the minimum value that separates two ‘instances’ (in our case, battles). It has to have a value of 1.0 or very bad things will happen.

 

So, let’s recap what we’ve got:

  • A series of algorithms that analyze a battlefield and return values representing various conditions that SMEs agree are significant (flanks, attack and retreat routes, unit strengths, etc., etc).
  • A Category Utility Function (CU) that uses the products of these algorithms to determine how similar analyzed battlefields are.

So now, we just need to put this all together. A battlefield (tactical situation) is analyzed by TIGER / MATE. It is ‘fed’ into the unsupervised machine learning function and, using the Category Utility Function one of four things happen:

  1. All the children of the parent node are evaluated using the CU function and the object (tactical situation)is added to an existing node with the best score.
  2. The object is placed in a new node all by itself.
  3. The two top-scoring nodes are combined into a single node and the new object is added to it.
  4. A node is divided into several nodes with the new objected to one of them.

These rules (above) construct a hierarchical tree structure. TIGER was fed 20 historical tactical situations (below):

  1. Kasserine Pass February 14,1943
  2. KasserinePass February 19, 1943
  3. Lake Trasimene, 217 BCE
  4. Shiloh Day 2
  5. Shiloh Day 1, 0900 hours
  6. Shiloh Day 1, 1200 hours
  7. Antietam 0600 hours
  8. Antietam 1630 hours
  9. Fredericksburg, December 10
  10. Fredericksburg, December 13
  11. Chancellorsville May 1
  12. Chancellorsville May 2
  13. Gazala
  14. Gettysburg, Day 1
  15. Gettysburg, Day 2
  16. Gettysburg, Day 3
  17. Sinai, June 5
  18. Waterloo, 1000 hours
  19. Waterloo, 1600 hours
  20. Waterloo, 1930 hours

In addition to these 20 historical tactical situations five hypothetical situations were created labeled A-E. This is the resulting tree which TIGER created:

The hierarchical tree created by TIGER from 20 historical and 5 hypothetical tactical situations. The numbers in the nodes refer to the above legend. Battles placed in the same nodes are considered very similar by TIGER. Click to enlarge.

If we look at the tree that TIGER constructed we can see that it placed Shiloh Day 1 0900 hours and Shiloh Day 1 1200 hours together in cluster C35. Indeed, as we look around the tree we observe that TIGER did a remarkable job of analyzing tactical situations and placing like with like. But, that’s easy for me to say, I wrote TIGER. My opinion doesn’t count. So we asked 23 SMEs which included:

  • 7 Professional Wargame Designers
  • 14 Active duty and retired U. S. Army officers including:
  • Colonel (Ret.) USMC infantry 5 combat tours, 3 advisory tours
  • Maj. USA. (SE Core) Project Leader, TCM-Virtual Training
  • Officer at TRADOC (U. S. Army Training and Doctrine Command)
  • West Point; Warfighting Simulation Center
  • Instructor, Dept of Tactics Command & General Staff College
  • PhD student at RMIT
  • Tactics Instructor at Kingston (Canada)

And in a blind survey asked them not what TIGER did but what they would do. For example:

Twenty-three SMEs were asked this question: is this hypothetical tactical situation (top) more like Kasserine Pass or Gettysburg?. Click to enlarge.

And this is how the responded:

Results from 23 SMEs answering the above question. Overwhelmingly the SMEs agreed that that the hypothetical tactical situation was most like the battle of Kasserine Pass.

So, 91.3% of SMEs agreed that the hypothetical tactical situation was more like Kasserine Pass than Gettysburg Day 1. Unbeknownst to the SMEs TIGER had already classified these three tactical situations like this:

How TIGER classified Kasserine Pass (1), Gettysburg Day 1 (14) and a hypothetical tactical situation (B). The cluster C1 contains two tactical situations that both have restricted avenues of attack caused by armor traveling through narrow mountainous passes. These passes also partially create restricted avenues of retreat. REDFOR does not have anchored flanks.Click to enlarge.

In conclusion: over the last four blog posts about Computational Military Reasoning we have demonstrated:

  • Algorithms for analyzing a battlefield (tactical situation).
  • Algorithms for implementing offensive maneuvers.
  • An Unsupervised Machine Learning system for classifying tactical situations and clustering like situations together. Furthermore, this system is never-ending and as it encounters new tactical situations it will continue this process which enables the AI to plan maneuvers based on previously observed and annotated situations.

This is the AI that will be used in General Staff. It is unique and revolutionary. No computer military simulation – either commercially available or any military simulation used by any of the world’s armies – employ an AI of this depth.

As always, please feel free to contact me directly with questions or comments. You can use our online email form here or write to me directly at Ezra [at] RiverviewAI.com.

References

References
↑1 TIGER: An Unsupervised Machine Learning Tactical Inference Generator which can be downloaded here
↑2 https://en.wikipedia.org/wiki/Turing_machine

Computational Military Reasoning (Tactical Artificial Intelligence) Part 3

1 Reply

In my two previous blog posts on the subject of battlefield analysis (computational military reasoning and tactical artificial intelligence) I discussed how the TIGER1)Tactical Inference Generator / MATE2)Machine Analysis of Tactical Environments program can identify certain key tactical positions such as anchored / unanchored flanks and restricted / unrestricted avenues of approach (attack) and retreat. In this blog post we will look at how TIGER / MATE performs analysis of frontages and implements the infiltration and penetration maneuvers.

MATE / TIGER employs a number of ‘building block’ algorithms that are used in various tactical and battlefield analysis algorithms. One set of building block algorithms are the Grouping algorithms that determine weaknesses in frontages. The following slides are from an unclassified briefing that I gave to DARPA (the Defense Advanced Research Project Agency) on my MATE program (funded by DARPA research grant W911NF-11-200024):

All slides can be enlarged by clicking on them. Note: OPFOR = Opposition Forces. In all of these slides OPFOR are the red units.

With the above building block algorithms we can calculate the optimum part of OPFOR’s frontage to set as the Schwerpunkt. Schwerpunkt is a German word meaning, “the point of maximum effort.” The term was used in blitzkrieg planning to specify, “the center of gravity, point of main effort, where a decisive result was to be achieved.”

Using the above algorithms we can now calculate the Schwerpunkt for implementing the penetration and infiltration maneuvers. The following slides are from an unclassified briefing that I gave to DARPA (the Defense Advanced Research Project Agency) on my MATE program (funded by DARPA research grant W911NF-11-200024): All slides can be enlarged by clicking on them.

The Five Canonical Offensive Maneuvers are from the U. S. Army Field Manual 3-21 which is available online. As always, if you have any questions please feel free to email me.

References

References
↑1 Tactical Inference Generator
↑2 Machine Analysis of Tactical Environments

Computational Military Reasoning (Tactical Artificial Intelligence) Part 2

Leave a reply

In my last blog post I described how the TIGER / MATE programs classified battles (in computer science terms ‘objects’) based on attributes and that anchored or unanchored flanks was one such attribute. After demonstrating the algorithm for calculating the presence or absence of anchored flanks we saw how the envelopment and turning tactical maneuvers were implemented. In this blog post we will look at another attribute: restricted avenues of attack and restricted avenues of retreat.

The only avenue of retreat from the Battle of First Bull Run back to Washington was over a narrow Stone Bridge. When a wagon overturned panic ensued. Library of Congress.

One classic example of a restricted avenue of retreat was the narrow stone bridge crossing Cub Run Creek which was the only eastern exit from the First Bull Run battlefield. The entire Union army would have to pass over this bridge as it fell back on Washington, D.C. When artillery fire caused a wagon to overturn and block the bridge, panic ensued.

At the battle of Antietam Burnside tried to force his entire corps over a narrow bridge to attack a Confederate position on the hill directly above. The bridge was famously carried by the 51st New York Infantry and 51st Pennsylvania Infantry who demanded restoration of their whiskey rations in return for this daring charge. From the original Edwin Forbes drawing. Click to enlarge.

Burnside’s Bridge at the battle of Antietam is a famous example of a restricted avenue of attack. Burnside was unaware that Snavely Ford was only 1.4 miles south of the stone bridge and allowed a back door into the Confederate position. Consequently, he continued to throw his corps across the bridge with disastrous results.

How to determine if there is a restricted line of attack or a restricted line of retreat on a battlefield

From the perspective of computer science restricted avenues of retreat and restricted avenues of attack are basically the same problem and can be solved with a similar algorithm.

As before we must first establish that there is agreement among Subject Matter Experts (SMEs) of the existence of – and the ability to quantify – the attributes of ‘Avenue of Attack’, ‘Avenue of Retreat’ and ‘Choke Point’.

The following slides are from an unclassified briefing that I gave to DARPA (the Defense Advanced Research Project Agency) on my MATE program (funded by DARPA research grant W911NF-11-200024):

All slides can be enlarged by clicking on them.

Now that we have determined if there is a restricted avenue of attack the next blog post will discuss what to do with this information; specifically the implementation of the infiltration and penetration offensive maneuvers.

As always, if you have any questions please feel free to email me.

References:

TIGER: An Unsupervised Machine Learning Tactical Inference Generator, Sidran, D. E. Download here.

Free Scenarios Twenty-One Through Twenty-Five

Leave a reply

We asked you for your Top 30 battles that you would like to see included free with General Staff for supporters of our Kickstarter campaign. We have previously announced the first twenty vote-getters. Today we are announcing the next five. One of the interesting features of General Staff is the ability to combine any two armies with a map to create a scenario. We use this feature for two day battles (such as Wagram and 2nd Bull Run) effectively creating two completely different battles (with two different armies) but using the same battlefield map.

This map of the battle of Alma was created only two years after the battle. Click to enlarge.

The battle of Alma is our first foray into the Crimean War. The Russians, though outnumbered, have the heights with their guns entrenched in heavy fortifications. The British and the French suffer numerous communication breakdowns. The battle seesawed back and forth until a final assault by the Highland Brigade carried the day and the Russians broke and fled from the battlefield. Playing the Allies will test your ability to coordinate attacks via messengers. Playing the Russians will require skillful coordination of counterattacks.

Wagram was a two day battle with the first day involving crossing the Danube. Click to enlarge.

On May 21st and 22nd Napoleon had attempted to cross the Danube at Lobau Island only to be turned back by Archduke Charles. Now, after over a month of preparations and reinforcements, Napoleon was ready to try again.

We present two distinct scenarios for the battle of Wagram: the first representing the situation on July 5th and Napoleon’s second attempt at crossing the Danube and establishing a beachhead and the second the battle of July 6th in which Archduke Charles attempted a double envelopment of Napoleon’s army. Only Napoleon’s hastily created ‘grand battery’ of artillery, a desperate cavalry charge and a counterattack by MacDonald’s corps saved the day. The Austrians eventually broke and fled the battlefield and sued for an armistice which ended the 1809 war.

Plan of the second Battle of Bull Run Va. Showing position of both armies at 7 p.m. 30th Aug. 1862. From the Library of Congress. Click to Enlarge

After General George McClellan’s disastrous Peninsula campaign, President Lincoln appointed Major General John Pope to lead the newly formed Army of Virginia and was tasked with the missions of protecting Washington D.C. and clearing the Shenandoah Valley of Confederates. McClellan, who never responded promptly to orders even in the best of circumstances, simply ignored commands to begin transferring his army from the peninsula southeast of Richmond up to Pope in front of Washington. Lee, knowing that McClellan had a bad case of the ‘slows’ took advantage of his interior lines to rapidly move his forces north to destroy Pope before McClellan’s troops could reinforce him.

The battle on the old Mananas battlefield began on August 28, 1862 with Jackson (commanding the left wing) shelling the passing Union column of King’s division (which included the soon to be famous Iron Brigade). The Iron Brigade, though outnumbered, attacked and fought Jackson’s famous division to a standstill. However, Jackson’s attack was primarily a feint employed as a ‘fixing force’ for an envelopment maneuver; Longstreet’s corps was expected to appear on the Union’s unprotected left flank.

On the second day, August 29th, Pope attempted to initiate a double envelopment against Jackson. However, Longstreet had now appeared on the battlefield at exactly the wrong place for Pope’s envelopment maneuver. The day was marked with incredibly poor communications between Pope and his subordinates and ended mostly as it began with neither side gaining or losing much ground.

The third day, August 30th, began with Longstreet’s counterattack on the Union’s exposed left flank. Again, incredibly poor communications between Pope and his subordinates turned a bad situation into a disaster. Unlike the first battle of Bull Run, the Union army fell back on Washington in an orderly column through an extremely limited avenue of retreat over Bull Run.