Monthly Archives: January 2016

Stop the presses! Go AI defeats human professional!

A few days ago, Google’s DeepMind announced that they reached the most significant milestone in pure AI since the solving of checkers in 2007: a Go AI called AlphaGo that’s competitive with human professionals. (Strictly speaking, computers could play Go before yesterday, but they were ranked around midlevel amateurs at their best.) The whole paper is available here.

For those of you who aren’t as much nerds about AI as I am, here’s a quick primer on why this was thought to be a very hard problem (so hard that the people involved in the prior state of the art thought it was at least a decade away):

In the most theoretical conception, game-playing computers for perfect-information zero-sum games (most abstract board games: those with no hidden state with all players working toward the same objectives, to be not entirely accurate but more readable than perfect accuracy allows for) are as simple as exploring every possible move and every possible countermove from the current position to the end of the game. Assuming perfect play on both sides, every result will be either a win, a loss, or a draw—that is, abstract strategy games are perfectly deterministic. (Checkers isn’t completely solved, but we do know now, as a result of the work from 2007, that perfect play on both sides from the start always yields a draw.)

This is, however, a massively impractical way to actually play a game, because the number of positions to explore rapidly turns intractable. Speed-focused modern chess engines search on the order of millions of nodes (in the game tree) per second, but searching a chess position exhaustively to a depth of 7 requires a search of about 60 billion nodes. Traditional games like chess and checkers yielded to some optimizations on this process:

  • It’s easy to look at a chess or checkers board and tell how well the game is going for a certain player: in both games, it comes down primarily to the balance of pieces. (The last I read, advantages in position are worth a pawn or two in a game of chess if they’re all going your way; the queen is worth nine pawns.) So, you don’t need to explore all the way to the bottom of the game tree to get an idea of which directions are promising. You just explore as deep as you can and evaluate the position.
  • If you have a good evaluation function (that is, one that generally only evaluates a position as better when it’s closer to winning), you can do some easy pruning when you come across a game state that’s known to be worse than the worst possibility you’ve explored: in that case, you just don’t explore any further in that direction. It works even better if you can assess which moves are likely to be good and explore those first: if you try the best move first, every subsequent move is going to turn out to be worse, and you’ll save a bunch of time.

So chess computers today, working with those tools, are better than the best human players: the effective branching factor (the number of moves to explore at each state in the tree), using the pruning techniques above, goes from about 35 to between 1 and 10, depending on the exact techniques used. The reason Go didn’t fall to traditional techniques is because it’s just so much more computationally difficult. Chess’s branching factor (the number of possible moves at each state) is about 35, and games last about 80 moves; Go’s branching factor is about 250 on average, and runs about 150 moves. It also features a few difficulties that chess does not:

  • It’s a very non-local game, both in space and in time: a move made at the very start of the game halfway across the board could have implications fifty turns later on the strength of the positions played at the start. This is a horizon problem: in chess, most positions become quiescent—not likely to affect the end effect of that position on the overall evaluation—after the captures stop. Modern chess engines will play all the way through capture sequences for this reason; there’s no similar metric to use for go engines.
  • It’s very difficult to evaluate a board position on purely objective grounds, or rather, we haven’t figured out how to phrase, mathematically, what about a good go position is good. Neither present control of territory nor present number of captures bears very strongly on the eventual result.

Because of the size of the problem space for Go, traditional techniques don’t work. The branching factor remains too high. Modern Go programs use one (or sometimes both) of two approaches: either they use hand-coded expert knowledge to sort and select promising moves for tree expansion (which frequently misses oddball moves that are nevertheless good), or they randomly play out a bunch of games from the current position to the end, and sample the result to pick the best move on aggregate (which frequently misses obviously good moves). The best of the pre-AlphaGo bunch used both, combining expert knowledge to pick the best moves to sample with the oddball-finding power of random simulation.

AlphaGo does a little bit of that, but at its heart, it learned to play a lot like humans do: DeepMind’s researchers fed it a diet of 30 million sample positions and the eventual results, and built a neural network to identify what a good board looks like and what it doesn’t. (Literally—the input into the program is a 19×19 image, with pixels set to values representing empty, black stone, or white stone.) They built a second neural network to identify which states are the best ones to simulate through in a random simulation, and Bob’s your uncle: a professional-level Go program. Their paper suspects it’s about as good as a mid-level human professional—the exhibition tournament they included in the paper saw AlphaGo beat the European human champ five games to zero, four by resignation, but the Euro champ isn’t a top-tier player worldwide. February will see an exhibition tournament between the South Korean champion, a world-class player; we’ll see how it does against him. AlphaGo also won 499 out of 500 games against the prior state of the art Go computers.

The most interesting thing about this development is that it learned to play a lot like a human would—it studied past games and figured out from that study what was good play and what wasn’t, then played a lot (against itself and past versions of itself), which is roughly analogous to playing a lot and working on go problems (the way human go players are supposed to get better). The obstacle to a general game-playing AI (I could buy a copy of Arkham Horror, finally, without having to rope the wife into an eight-hour marathon of doom!) is that training neural networks is currently pretty slow. As I mentioned in the first post, AlphaGo had to study about thirty million positions and play itself many, many times to get to the level of competence it’s at now; presumably, that will improve as DeepMind hones its learning techniques and discovers new and faster ways.

That said, humans still have one thing to be proud of: efficiency. The version of AlphaGo that beat the European champ ran on about 1200 CPUs and about 200 GPUs, whereas the human brain, which is nearly as good, draws about 20 watts.

Some extra reading for you: GoGameGuru has all the game replays, which are interesting if you’re familiar with Go (it doesn’t take much—if you’ve seen a human game or five and a game-against-computer or five, you’ve probably already noticed the differences in play, if only subconsciously) AlphaGo has a style that looks very human. Apparently, Go professionals expect the Korean champ to beat AlphaGo, and another source has the game taking place in March. If I were the champ, I’d be pushing for, like, next week. I don’t know how long the million-dollar prize is gonna be good for.

Here’s a commentary on one of AlphaGo’s games against the European champ.

I originally wrote this analysis for the people over at the Quarter to Three forum, which explains the breezy, off-the-cuff style. If you’ve been following us for a while, some of the introduction will have seemed repetitive, too. Anyway, I hope it was still informative and interesting. -Fishbreath

It takes two to tango: why I like single-seat attack helos

Picture your favorite helicopter gunship. I can’t tell you much about it without knowing what it is, but I can tell you one thing: unless you’re a weirdo like me, it has two seats. I do not think this must be so. To explain why is going to take a little detour into the tactical thinking of helicopter pilots, and how that affects the way they’re employed on the battlefield.

Picture yourself as a fixed-wing pilot. You can easily fly above all but the most specialized of ground-based weapons systems. Compared to anything in the dirt, you are extremely fast, so fast that they may as well be standing still. Your bog-standard general purpose bomb is several times more explosive than the largest explosive projectiles commonly hurled by things on the ground. Your precision-guided weapons are more precise, your sensors are better, you can see further. You are as unto a god, or at least a hero of Greek or Norse myth, striking down your foes with the weight of inevitability behind you.

Got that image in your mind? Savor it for a minute. Now forget all about it, because that isn’t how flying a helicopter works at all.

Picture yourself as a helicopter pilot. If you fly high, a plane will shoot you down, or a long-range air defense system. If you fly low, things on the ground a plane would laugh at will shoot at you, and might shoot you down. You are fast, but you aren’t so fast that you can really use it to enhance your survivability. You do not generally carry especially heavy weapons, and your sensors are pretty good, but you aren’t high enough to see a long way. You are certainly not as unto a god. You’re scary, but it’s the kind of scary your adversaries can actually kill.

What does that mean for you, noble helo pilot? How does it shape your doctrine? If you’re looking for a metaphor, the right analogue for a helicopter is not an IFV or a tank. If you’re a helicopter pilot, your mindset is ‘sky infantry’. You keep out of sight, use natural cover, engage quickly before getting out of sight, and generally skulk around in the mud. Just like the infantryman has a pretty bum deal on the ground, the helo pilot has a pretty bum deal in the sky. The only difference is that the helo pilot has someone to look down on.

Why do attack helicopters generally feature two crew? Because there are three jobs in a helicopter, and one person can’t do all three at once. You need to fly the helicopter, which is a difficult task on its own; you need to use the weapons, which often requires going heads-down; you need to keep your eyes up to see threats visually, since a lot of the things that can shoot down a helicopter can only be detected by the Mark I Eyeball1. The pilot can fly and watch, if the gunner is working with the sensors or weapons systems, and the gunner can keep an eye out, if the flying gets especially hard on the pilot. Simply put, each crewman can do about one and a half things simultaneously, and each helicopter has three things you need to do. Perfect coverage.

Mathematically, it looks bad for the single-seat concept. One crewman can do one and a half things. The helicopter has three things that need to be done. Let’s work on bringing those numbers closer together.

First off: we can install an advanced autopilot. We’ll go the Ka-50, the only single-seat attack helicopter ever to see combat service, as our example2. Taking its age into consideration, the Ka-50 has one of the most advanced autopilot systems ever installed in a helicopter. It’s fully capable of flying the helicopter through a noncombat mission from just after takeoff to just before landing, and can take control in nearly every combat situation that doesn’t involve immediate evasive action, or nap of the earth flying. This reduces our list of things to do to two, but we still only have one and a half tasks doable with our single crewman.

How can we fix that? Add a second crewman, but put him in a different airframe. Your helicopters fly in pairs. How many things will we need to do at once? Fly, but the autopilot takes care of that for us. Use weapons, yes, but that’s a shared task: only one helicopter needs to be engaging at a time. That’s one thing between us. Keep an eye out, yes: ideally, both of us should be keeping an eye out, but in a pinch, one pilot can keep an eye out for the whole team. That leaves us two crewman, who together can do three things, and two or three things to do between them (that is, weapons, eyes, eyes, or weapons, eyes).

That’s really all there is to the argument. Additional automation can help reduce the workload further. A fancy threat warning system helps reduce the need for constant lookout, and helps direct pilot attention during the few, emergency situations where the autopilot is insufficient. Better weapons and datalinks allow for off-board targeting, which can be used to move the weapons employment burden around between helicopters. Autopilots with more options yield further reductions in flying workload—a terrain-following radar or lidar, for instance, would give the Ka-50 the ability to fly nap of the earth at high speeds. Better sensors help reduce the time spent heads-down for weapons employment.

I’m nearing my target word count here, so I’ll wrap up with some quick pros and cons. I’ve made a decent argument that a single-seat attack helicopter is a reasonable choice, so why might you prefer one? To start, you have reduced aircrew requirements, and reduced aircrew losses—half of two airframes is one, and half of one airframe is zero. You have a great deal of large-scale tactical flexibility. Since the two-ship element is the basic unit of maneuver, you can choose to advance in bounding overwatch, for instance, or widely separate your eyes from your weapons. Your eyes helo might be just behind solid cover on a ridge outside of enemy engagement range, able to peek and feed coordinates to your weapons helicopter, which might be advancing in concealment much nearer the enemy. In separating eyes and weapons, terrain may sometimes allow a quick attack from two angles in rapid succession, or at entirely the same time. If you have a small number of helicopter pilots, single-seat airframes let you put more into the sky at once. It’s a setup optimized for tankbusting: large targets, relatively easily spotted and shared.

Why might you choose the standard two-seater? It’s better in moderately threat-heavy COIN situations, where the front lines are poorly defined and any territory may become enemy territory. Two-seat helicopters have better small-scale tactical flexibility, and a single two-seat helicopter swing between navigation, evasion, and counterattack much more quickly than a pair of single-seat airframes. For another, two-seaters are tried and tested. Nobody operates a single-seat attack helicopter in any real number today, not because it’s not a workable theory, but because the only modern example entered service well after its technology started down the hill toward obsolescence. Today, you’d have to build your own single-seater, or buy a bunch of Kamovs and refit them, while you can buy Havocs or Cobras or, for that matter, the Ka-52, basically off-the-shelf. Two-seat helicopters have better engagement speed: for a given number of helicopters and a given number of weapons, the two-seaters will distribute their arms faster, because each airframe is a self-contained targeting and shooting unit, not depending on another helicopter for overwatch or targeting data.

That’s about all I have. One of these days, I’ll take a look at the concept, and come up with some justifications for why Luchtburg might choose a single-seat helo.

1. Or the Mark II Eyeball, also known as the missile launch warning system.
2. The Ka-50 is outmoded in today’s market, but if you look at its competitors in late 80s, when it first appeared on the scene, it’s a much closer case, and depends mainly upon some tactical considerations I’ll get into later.

On tafl: OpenTafl Notation

I mentioned in a previous post that one of the practical difficulties involved in computer implementations of tafl is the lack of a support infrastructure: anyone desiring to write a tafl AI must also write a tafl engine. Admittedly, this probably isn’t a significant obstacle to anyone with a real desire to write a tafl AI, but it does yield a great deal of duplication of effort. This is why my next step with OpenTafl is to build it into an engine into which other AIs can be plugged, and my eventual aim, hopefully at the end of this year, is to host the world’s first computer tafl tournament, where all the AIs so far written can compete against one another for bragging rights1. The timing depends mainly on how quickly I can write the features I need, but I hope to run the 1st Many Words Computer Tafl Open in late December2.

To do this, though, three sorts of notation are required. First, a method for communicating positions; second, a method for communicating rules; and third, a method for communicating moves. (A method for communicating moves already exists, thanks to tafl historian Damian Walker, but I propose some enhancements to it later.) The full version of my proposal is included at the end of this post.

First off, a method for positions: I propose something akin to chess’s Forsythe-Edwards notation, which is easily machine-readable, but also human-readable, and easy to compose by hand. Here’s an example, representing the brandub starting position:


Enclosed by slashes are rows. Lowercase letters represent pieces from the attacking side. Uppercase letters represent pieces from the defending side. (Besides t: taflman and k: king, my proposal also includes n: knight and c: commander, for Berserk-variant tafl.) Numbers represent that many empty spaces.

To communicate rules, there’s very little choice but to do so exhaustively. Clever use of defaults can somewhat alleviate the issue. Here is a proposed OTN rules string for Fetlar tafl:

dim:11 atkf:n start:/3ttttt3/5t5/11/t4T4t/t3TTT3t/tt1TTKTT1tt/t3TTT3t/t4T4t/11/5t5/3ttttt3/

It uses the dimension element (dim:11) and the start element (using an OTN position string) to mark the start and end of the rules; other elements come in between. The defaults suggested in the proposal, with the exception of atkf (attackers move first), are the rules of Fetlar, and other tafl games may be constructed as transformations from Fetlar. For instance, here’s brandub again:

dim:7 ks:n cenhe:n cenh: cenp:tcnkTCNK start:/3t3/3t3/3T3/ttTKTtt/3T3/3t3/3t3/

Brandub is Fetlar, except it’s played on a 7×7 board (dim:7), the king is not strong (ks:n), the center does not become hostile to the defenders when empty (cenhe:n), the center is not hostile to anybody (cenh:), and pieces of all types may pass over the center space (cenp:tcnkTCNK). Here’s Sea Battle 9×9:

dim:9 esc:e ka:n cenh: cenhe:n corh: cenp:tcnkTCNK corp:tcnkTCNK cens:tcnkTCNK cors:tcnkTCNK start:/3ttt3/4t4/4T4/t3T3t/ttTTKTTtt/t3T3t/4T4/4t4/3ttt3/

Sea Battle is Fetlar, except it’s played on a 9×9 board, escape is to the edge (esc:e), the king is not armed (ka:n), the center and corners are hostile to nobody, all pieces may move through the center and the corners, and all pieces may stop on the center and the corners.

Of tafl variants of which I am aware, the options in the rules string are sufficient to support all of them but one very esoteric alea evangelii variant3, and OpenTafl can theoretically support any variant constructed using OTN rules strings4. Other notable capabilities include the ability to define certain spaces as attacker fortresses, to support hostile attacker camps.

Finally, we move on to move notation. Building on Damian Walker’s aforementioned, excellent adaptation of algebraic chess notation to tafl rules (which, I should stress, is more than sufficient), I wanted to create a notation which requires a minimum of inference, and therefore slightly better suited to inter-computer communication. In effect, I wanted a notation sufficiently detailed such that a computer, operating with only a starting position and a list of moves, could accurately replay the game without having to know any of the rules. Since it is a notation with a lot of options, I provide a deeply contrived sample move record here, which uses almost every option:

Ne4-e6xf6 Ne6^=e8xe7/e9/cf8/d8-

The defending knight (uppercase N) on space e4 moves to e6, capturing the piece on f6 (having no special letter, it’s just an ordinary taflman). This is berserk tafl, so the move is not yet over. The knight on e6 jumps to e8 as its berserk move (^: jump, =: berserk), capturing the taflmen on e7, e9, and d8, and the enemy commander on f8 (as a special piece, it gets a prefix; as an attacking piece, it’s lowercase). This opens up a path for the king to reach the edge (-: king has a way to the edge).

You can read more, and peruse more examples, in the full proposal here: OpenTafl Notation. Feel free to leave a comment with your, well, comments.

1. And maybe a $20 Amazon gift card or something.
2. I hope to finalize the protocol in the next few months, and finish implementing it by the end of summer.
3. The one with mercenaries, pieces which start on the defending side but change sides when captured.
4. I haven’t yet written the code to parse them and generate new games, so I can’t say for sure. Thinking through some oddball variations, though, none strike me as OpenTafl-breaking.

Happy Birthday, John Moses Browning!

Happy Birthday, John Moses Browning!

He’d be 161 years old if he were alive today. You might know him as the designer of the M1911 handgun, which is dear to my heart. You might also know him as the designer of the Browning Hi-Power, the first double-stack mag 9mm pistol. It’s the first Wondernine, about fifty years before anyone started using the term. I’m not super fond of the Hi-Power, but that’s ok. The double stack handgun magazine idea has persisted into guns that I’m more fond of. Like my Glocks.

But John Moses Browning had many more ideas that you may or may not realize belong to him. He first came up with the tilting barrel lockup system, which is now used on just about every handgun1 in production. It beat out all other designs for the first choice because it works well and it’s cheap and easy to machine. It also doesn’t need a ton of lubrication, unlike most alternatives.

John Moses Browning also patented the notion of a slide on a self-loading pistol, and all of them have that these days. It’s just so darn useful. You can look at a lot of goofy pistol designs of the early twentieth century and see designers trying to work around this patent, and the ideas went nowhere once that patent expired.

But wait, there’s more! John Moses Browning’s first commercially-successful handgun design, the FN Model 1899, was striker fired. Yes, just like the Glock 17 currently sitting on my hip. Good old Browning. Getting ahead of the game there. Even though people of the time thought it was exceedingly odd for a pistol to not have a hammer.

John Moses Browning also designed a rather solid machine gun. The rifle caliber incarnation was lighter than the maxim and plenty reliable, and was quite popular as an aircraft weapon. The .50 BMG version, the venerable Ma Deuce, is still in service today. Introduced in the ’30s, it’s been used on just about every vehicle you can think of, from World War 2 fighters to tanks, to helicopters, to ships, to jeeps. If it’s a vehicle, it’s probably had a Ma Deuce on it. It’s even been used as a sniper rifle. Carlos Hathcock got a kill at 2,250 yards with one. Which is pretty impressive for a crew-served, tripod-mounted weapon.

John Moses Browning developed a ton of other stuff too. I could go on, but the above are the biggest things that are still taken advantage of today. Happy Birthday, John, and many happy returns. We certainly have gotten many such returns from you.

1.) Or at least all the popular, cool ones.

Milspec Challenge

Here’s one that came from Fishbreath. I’m a big AR-15 guy, and he suggested I spec one out as if I was going to issue it. So here goes. An issue AR-151, Parvusimperator-style.

We’ll start with the receivers. Both upper and lower receivers should be forged from 7075-T6 aluminum. Forged to keep it simple. The forging process is cheaper than the billet manufacturing process, and marginally stronger. Win-win. We’ll focus on the lower receiver and its components first.

The trigger group is a safe/semiautomatic/fully-automatic one. The AR-15 trigger group allows for a three-position selector without some serious modifications. The reader might wonder, ‘why not three round burst?’ My response is ‘why bother?’ The three round burst is the latest in a long line of devices designed at the urging of girly-man bean counters to force soldiers to not “waste ammo”. In the 1870s, the US Army didn’t issue the Winchester repeating rifles to the cavalry because the soldiers might waste ammo. Bolt action rifles around the turn of the 20th century had a magazine cut off, because soldiers might waste magazine after magazine of bullets. Both decisions were proven wrong. Discipline must be taught; it cannot be enforced by weapons. You might not always need ammunition, or fully automatic fire, but when you do, you really need it. So safe/semi/auto is the trigger group we’ll use. We’ll also specify an ambidextrous selector switch.

The receiver extension should be 1.14 inches in diameter, and should have six position detents on the underside. This part may also be called the “buffer tube” but it’s better to think of it as an extension to the receiver that gives more room for the bolt carrier to decelerate. It should be held in place by properly-staked castle nut. This will ensure that the castle nut will not back out on it’s own, but can be removed by an armorer if necessary. In terms of buffer internals, we’ll use chrome silicon wire springs (something like the Springco Blue model extra power spring) and an H32 weight buffer.

As for the stock, we’re going for the Crane3 SOCOM-pattern stock. This is heavier and sturdier than the standard collapsible stock, and comes with two waterproof storage compartments. These are perfect for storing batteries for the various issue electrics that will get attached to the carbine, as well as small bottles of lubricant to keep the gun running. It’s a convenient place to store some lubricant to make sure it’s available when needed.

There are a few other gubbins we need to specify. We’ll specify an ambi bolt catch and an ambi mag catch. Installing these in a convenient manner will take some reworking of the lower, but it can be done. Knight’s Armament and several of the fancy billet receiver companies have already done this. I’m not overly particular about most other small parts. Appropriate aluminum pins, etc. are fine. The one other thing to call out is the pistol grip. I hate the A2-type grip, so that’s right out. It’s too small and it has a pointless nubbin in a stupid place. Fortunately, there are lots of alternatives out there, and almost all of them are better. I’m a fan of the TangoDown BG-16 grip, as it fits the hand better and doesn’t have a stupid nubbin on the front in the wrong place, so let’s go with that.

That takes care of the lower. Onto the upper. We’re going to call for a 14.5″ barrel. Why 14.5″? Because it’s a military standard with lots of available data, and it’s shorter and handier than a 20″. This is a general issue carbine, so barrel length is going to be a compromise. Plus, that hot deathray M855A1 ammunition is designed around a 14.5″ barrel. It will have a 1:7 twist rate4 and a lightweight profile. Why a lightweight profile? Because I don’t see a point to anything heavier. A lightweight barrel was perfect for full-auto ambush drills in Vietnam. A lightweight barrel will take something like 900 rounds before it fails, and that’s if you reload as fast as you can and never stop firing. That’s about three times a basic ammo load for the average soldier. This is more than adequate for a select fire carbine. It’s not a support weapon built for sustained automatic fire. The barrel should be made of 4150 chrome molybdenum steel. And yes, Virginia, it will have a chrome lining.

The gas system should be a carbine-length one, or about seven inches from the receiver to the gas block. Again, this is the standard on a 14.5″ barreled gun. The gas block should be a low-profile unit held onto the barrel with a pair of taper pins. This is the most secure method of securing a gas block to a barrel. It should also have a bayonet lug at the front, since the length is right for a bayonet. And bayonet fighting teaches a lot of good things to troops, plus it’s occasionally very useful on the battlefield. Just ask the British. For they bayonet lug to work, the gas block has to be as long as a front sight block, but this isn’t hard to do.

For handguard, we actually want a 9″ long unit, even though our specified gas system is only 7″ long. This will cover most of our low profile gas block, giving us more room for a grip and accessories, but still let us access the bayonet lug. We want a quadrail handguard, because picatinny rails are the standard accessory mount. There are some alternatives out there, namely keymod and mlok, but they don’t totally replace the picatinny rail. We see no reason to adopt two mounting systems, so we’ll stick with the picatinny. I went back and forth quite a bit on which handguard design to call out here. I’m rather fond of the notion of a monolithic design, where the rail and upper are one continuous piece. However, this is generally heavy, and doesn’t let us change the handguard out at all. One of the joys of the AR design is just how modular it is. We’re using a very different handguard design than my father would have found on an AR-15, and we’d like our sons to be able to apply their own wisdom. So let’s skip the monolith, cool though it might be. We want a durable design though, and since we’re covering our gas block, we’ll need something tube-style, i.e. free float. No mall ninjas, we’re not getting a free floated barrel for accuracy benefits. It’s a service carbine. Don’t kid yourself–no one cares about tiny group improvements. We just need it to work with our gas block system. We’ll take the Daniel Defense DDM4 rail, in the 9″ length. This is an off the shelf product, just like everything else on our list. It’s lightweight, sturdy, and easy to install. It will still let us access the bayonet lug, but give us plenty of room for lights and lasers as needed. It also has quick detach sling swivel sockets built in, which is convenient. Saves us a part.

So far, so good. There are a few more notes that we’ll make for the upper. The muzzle device should be a three-pronged-style flash hider, since these seem to work the best. It should also be a suppressor adapter. Since we haven’t specified a suppressor, we won’t specify a muzzle device. But a good example would be the Surefire SF3P, which would work with Surefire’s excellent line of suppressors.

You may have guessed we’ll be using a “flat-top” or “A3-style” upper receiver, and you’d be right. That’s prime optic mounting real estate. And we’ll have the standard folding dust cover on there too. Yawn. Tell me something interesting, you say? Fine. We’re specifying our uppers to not have a forward assist. Why? Because it’s a stupid, useless, protruding, weight-adding piece of junk. Get rid of it. It makes the design simpler. It makes the design lighter. It’s how Stoner originally intended it to be. And stupid Army was wrong to insist on such a device. If you’re in a situation where you need to force the bolt closed, then your gun is FUBAR, and unsafe to use. Any case so screwy that you have to force it into the chamber because the buffer spring won’t do the job has no business in your gun. Yes, you can force a fucked-up, bulged case into an M1903, M1 Garand, or M14, but that doesn’t make it a good idea. Or even a military requirement. The M1918 BAR has no such capability, but troops loved the BAR all the same. Look, when both John Moses Browning and Gene M. Stoner don’t think it’s needed, it’s not needed.

Didn’t see that coming, did you?

Now for the guts of the rifle. But first, the charging handle. Surprise, the stock one is fine. It doesn’t need ambi anything. It’s already ambi. It’s on top of the freaking gun, for crying out loud. Fine as is.

Anyway, the guts. Or, more specifically, the bolt carrier group. We’re going to call out somewhat nonstandard parts here. No, we’re not going to retrofit a piston. There’s already one there. It’s inside the carrier, you dumb idiot. We’re going to use a specialty bolt carrier and bolt, specifically the Lewis Machine and Tool Enhanced Bolt Carrier and Enhanced Bolt. Why? Well, let’s take a look at our system, here. We have a shorter gas system length than Stoner originally specified, so we’re getting more gas pressure than Stoner originally designed around. The M193 ammunition that was originally used had an average chamber pressure of 52,000 psi. But we’re calling for something like M855A1, which takes advantage of more consistent modern powders to increase the average chamber pressure to 63,000 psi, which of course means more gas pressure for the system. Plus, we like suppressors, which means yet more gas pressure. What does all this gas pressure mean? Well, it means we’ve got more velocity on the bolt, which increases the likelihood that the bolt will try to unlock to early, which puts a shear and bending load on the bolt lugs. Surprise, this is bad for bolt life. The Enhanced Bolt Carrier has more vents and a revised cam path to reduce pressure and slow unlocking of the bolt. The revised bolt has a different extractor, using two springs instead of one, which is designed to give longer spring life. The lugs are redesigned for added strength as well. The end result is a longer-lasting system. Plus it’s a slightly not-stock part, so I’m not just picking a rifle off the rack here and calling it good.

So how is the rifle issued? Well, it comes with an issued optic, which we haven’t chosen yet. But there will be an optic. There will be a two-point, quick-adjust sling. More convenient than a standard carry strap type sling that your grandfather had on his M1 Garand. There are three rail panels issued. I prefer the Tango Down units, but the exact model isn’t very important. Just something so that the picatinny rail doesn’t cut your hands up after lots of campaigning. The rifle is also issued with a set of folding back up iron sights. Specifically the Troy industries M4-style front and their standard (non-dioptic) rear sight. The rear sight is “A1 style”, i.e. it’s has two apertures, but is not adjustable for elevation. Nor should it be. It should be durable and simple. It should be zeroed and kept safely in reserve. It should not be complicated. It is a backup sight on a carbine, not iron sights on a National Match rifle. Keep it simple.

The carbine is issued with iron sights and optic fitted and properly zeroed. Since we’re not dumb, our weapons racks are designed to hold the carbines with optics mounted, so soldiers can trust that the carbine they draw from storage has a good zero.

Some of you may be wondering what I would do if you made me get an AR-15 that I couldn’t piece together as above. What if you had to buy something off a shelf, all-up, Parvusimperator? Simple. I’d call Knight’s Armament Corporation, and ask if they give a bulk discount on large carbine orders. So there. Smartass.

1.) I’ll be using AR-15 as the generic term here, so as not to sound like I’m specifying brand or model details. Also, the AR-15 designation came first.
2.) 5.6 oz weight.
3.) NSWC Crane designed, produced by LMT and B5 Systems.
4.) I.e. one complete rotation in seven inches of travel

The Crossbox Podcast: Episode 3

This time on the Crossbox Podcast, we discuss Wargame: AirLand Battle, some indefensible positions, and our arms choices for a sadly underrepresented type of competitive shooting challenge.


Recommended reading:

Fishbreath Plays: SimplePlanes

I’m a fan of sandboxes.

Many of my favorite games are sandboxes, or have a sandbox element: StarMade is altogether a sandbox, Rule the Waves gives you plenty of latitude to build your own navy and your own history, Falcon 4 lets you play with someone else’s castle or kick it down as you please, and Command Ops, though less open than the rest of this list, still gives you the chance to do largely as you please inside whatever scenarios you can find or make.

So, when I saw that SimplePlanes, an aeronautics sandbox by Jundroo, who made one of my favorite physics puzzle games on Android, was now on Steam, I had to give it a whirl. We’ll get the bad out of the way first: it’s a port of a mobile game, and so the interface is not as powerful as, say, Kerbal Space Program’s (which is the natural comparison), and the parts list isn’t quite as lengthy as I’d like. That said, the flight modeling is excellent for a wee building game like this, and as with any building game, there are some superb examples out there. For another downside, there isn’t a lot to do; as far as I can tell, there isn’t a way to add new maps or new challenges, which is a shame. Either one would add a ton of longevity to the game. Finally, the combat bits could be expanded upon a little—damage is very binary right now, and hitting a plane with anything will usually pop it.

With that out of the way, let’s talk about the good. I’m going to do this by discussing some of the things I have built; namely, the aircraft carried by the zeppelin Inconstant, from Skypirates: the Kestrel, Falcon, Vulture, Albatross, and Gorcrow. All are based off of real-world designs. The Kestrel is a riff on the XP-55 Ascender, the Falcon is based on any number of (generally French) twin-boom pusher designs of the immediate prewar and postwar periods, the Vulture is a recreation of the Sh-Tandem, a Russian ground-attack design, the Albatross is a Blohm & Voss-inspired asymmetric design, and the Gorcrow is more or less every medium bomber between 1930 and 1945. (Note that I made a few modifications to fit my zeppelin-borne aircraft requirements and restrictions, which you’ll find at the end of this post.)

The Kestrel is one of my favorites, owing to its handling characteristics. The twin coaxial engines, with a total of 1,500 horsepower for only 6,000 pounds full load, push it to speeds in excess of 400 miles per hour. It fields an excellent anti-aircraft punch, and has superb maneuverability at high speeds. Its weakness comes mainly in its low-speed handling: its vertical stabilizers are small, to limit the drag they add, but this creates a prominent tendency to yaw instability at landing speed. As such, it’s a design that’s likely very prone to landing mishaps, and requires a steady hand on the stick and active feet on the pedals to put onto the skyhook. Though the design is unusual, it flies very well, responding smoothly with little adverse yaw or other undesirable handling characteristics. At the edges of its envelope, it can sometimes get the pilot into trouble; unrecoverable flat spins are a possibility.

In design, the Falcon is much more conservative: it treads on no unusual aeronautical ground. The twin-boom design provides some added damage resistance; losing the back of one boom isn’t immediately fatal. It’s powered by a 1,250-horsepower engine, about the largest single engine we can expect to find in the world of Skypirates, and has a maximum takeoff weight of about 9,000 pounds. (The version posted is overweight, and needs to be slimmed down.) With rather a lower power-to-weight ratio, it only reaches about 320 miles per hour, significantly slower than the Kestrel. Although its gun armament is less heavy than the Kestrel’s, it makes up for that loss in firepower by mounting several racks for air-to-air and air-to-ground rockets. Its flight characteristics befits its character: rugged and dependable, with very few surprises, although it does have a tendency to stall the lower wing in tight, low-speed turns.

The Vulture is probably the one whose looks most closely match its intended purpose. A light bomber and ground-attack plane, the Vulture is the usual aircraft of choice when Inconstant needs to make a show of force. Its unusual design gives it a great deal of lift for its footprint, and permit all of its hardpoints to be placed along the same axis as its center of mass: dropping weapons doesn’t change its balance at all, making it a forgiving platform when carrying large weapons. The centerline mount supports an aerial torpedo, but only when the plane is air-launched—aerial torpedoes are too long otherwise. (Note that Inconstant doesn’t carry Vultures equipped with landing gear.) To my surprise, the Vulture’s handling is docile in the extreme, even when fully loaded, and turns downright peppy when empty, even though it only sports a 1,000-horsepower engine. I ran into no surprises anywhere in the envelope.

The Gorcrow, powered by a pair of 700-horsepower engines, is a conventional medium bomber, with all that implies. Its handling is ponderous, but it can sling a heavy load of bombs or rockets, or three aerial torpedoes, making it Inconstant‘s heaviest hitter by far. Three gun positions, one at the back of each engine nacelle, and one atop the fuselage, round out its weapon fit. Again, an unsurprising performer—not spritely, and predictable in its handling. Unlike the other aircraft on the list so far, its bringback weight is somewhat less than its full fuel empty weight. Inconstant being fairly light on avgas stores, her Gorcrows are generally only launched when absolutely necessary, to avoid having to dump fuel overboard before landing. The in-universe version has a glazed nose, but I haven’t figured that out yet.

The Albatross, powered by two 800-horsepower engines, is a long-range transport aircraft, and also one of my favorites for its sheer unlikeliness. Although Herrs Blohm und Voss built similar aircraft for the Luftwaffe during the Second World War, I was a little concerned that the flight engine wouldn’t handle it well, given the presumably-complicated aerodynamics at play. To my surprise, it worked fine, and isn’t even particularly touchy. Anyway, the 1,600 combined horsepower pushes her to a good turn of speed when empty, nearly as fast as the Falcon, and pegs her total cargo capacity at just over four tons. The asymmetry does mean she has some slight balance concerns, but in-universe, it’s easily trimmable. Low-speed handling is good, thanks to the fat wings. Even with the asymmetric nature of the pitching and yawing forces, owing to the offset position of the empennage, it has surprising maneuverability when empty. Same remark about the glazed nose.

Now, I didn’t even get into the built-in challenges, or into serious modding. I was just messing around, and in the course of learning how to build airplanes, building these, and coming up with my flight reports, I got more than my $10 of fun. I also got at least $10 of storytelling value out of it: I now have quirks and flight characteristics in mind better for each of these planes than I did before, and I can work that into stories.

If you’re looking for a plane construction sandbox, look no further.

Fishbreath’s Zeppelin-Borne Aircraft Construction Rules for SimplePlanes

  1. Airframes should range between about 3 tons and 12.5 tons full load.
  2. Aircraft must be shorter than 70 feet and have a wingspan less than 110 feet.
  3. No single engine may develop more than 1250 horsepower.
  4. Aircraft must have a bringback-weight stall speed of 110mph or less. (The other 20-30mph to get down to zeppelin speed is assumed to come from flaps.)

Squad Support Weapons

Ack, the board has become overrun with Taflmen. Thanks a lot, Fishbreath. Fortunately for you, dear reader, I have just the solution:

Underneath our starry flag, Civilize ’em with a Krag!
And return us to our own beloved home.

Wait, no, wrong century. Krags are rather old fashioned. What about more modern weapons? Well, we’ve already opted for the kickass HK 416 for general issue carbine. Let’s look at what our soldiers have in terms of lightish support weapons.

We’ve established an eight man dismount squad, nominally comprised of two fireteams of four. Let’s talk basis of issue of some support weapons. We’ll also flesh out a few more choices.

We’ve established our choices of rocket launchers and anti-tank missiles in the Panzerfaust 3 and the FGM-148 Javelin, and will not repeat the reasoning here. In any case, we’ll call for one Javelin CLU and one PzF 3 computerized Dynarange sighting unit per squad. In general, we’d expect two Javelin missiles and two Panzerfaust 3 rockets, some combination of the PzF 3T anti-tank rocket and the PzF 3B demolition rocket. This is some of the joy of a mechanized force: the IFV can carry the weapons when you don’t need them.

We’ll also allocate each fireteam one underbarrel grenade launcher. This needn’t be carried all the time, but it’s a quick way to give the squad some indirect firepower. For underbarrel grenade launcher, we want something that needs nothing beyond a picatinny rail for a mounting solution. Our choice is the HK AG36. It’s a 40mm grenade launcher that is as modular as we need, plus it has a sideways-opening break action that lets it accomodate longer specialty 40mm rounds. And yes, it attaches to picatinny rails.

You’re probably waiting for the machine gun choice though. Every squad needs a machine gun. Or two. Or three, even. We’re going with two, one per fireteam. Again, nothing surprising here. Now we ought to choose a machine gun. There are many to choose from, but first we ought to determine the caliber. Nato-wise, we’ve got 7.62x51mm and 5.56x45mm. All of the advantages of the smaller, lighter round are apparent. Troops carry more, full stop. But this is a mechanized army, and this brings a wrinkle. The CV90, like nearly all other self-respecting IFVs, has a 7.62mm coaxial machine gun, and provision for several hundred rounds of 7.62mm ammunition in belts. The infantry, of course, have 5.56mm carbines, but those operate with “loose”1 5.56mm ammunition in magazines. So, with a 5.56mm squad automatic weapon, we have three kinds of small arms cartridges to ship, according to the logistics tables: 5.56mm magazines, 5.56mm link, and 7.62mm link. If we go with 7.62mm, we drop this down to two kinds, plus the infantry can share ammo with the coax. One big pool of reserve ammo. So we’ll go with this.

Now, the question becomes, which 7.62mm NATO machine gun to choose? Here, weight becomes an important factor. There are lots of excellent, big, heavy machine guns out there, built for lots of abuse and sustained fire. Chief among them is the excellent FN MAG 58.2 This is our coaxial- and pintle-mounted gun of choice, but it weighs 11.79 kg (about 26 lbs). Nicely accessorized with modern rails and a heat shield in the M240B variant brings us up to 12.5 kg (27.6 lbs). Both weights are unloaded and without optics, and we can see how it’s hard for a MAG gunner to move with his squadmates. He can’t assault positions very well. There is a lightened version available, the M240L, which uses expensive titanium parts to cut weight, and manages to get under 10 kg (21.8 lb) if you reduce the barrel length to about 20″ and use a collapsible stock.

Can we do better? Sure, provided we give up some durability for lots of sustained fire. And this is ok for our needs. Remember, this is for the squad automatic weapon. It’s not for sustained fire from a vehicle or a weighted tripod. It’s for support of the close attack. Fire and movement. Plus, the IFV is going to be a better base of fire anyway. The Russians have a superlight machine gun in the PKM, which comes in at a svelte 7.5 kg (16.5 lbs). But it’s in 7.62x54mm R. The R is for Rimmed, and while archaic, the brilliant Mikhail Kalashnikov used it to his advantage in the feeding mechanism. One might think to convert it to 7.62x51mm Nato, but then we gain wait. The Poles have done this in the UKM-2000, and that weighs almost a full kilo more, coming in at 8.4 kg (18.5 lbs). Oof. That’s still lighter than the M240L though.

We might think to try the M60, but that comes in around 10.5 kg as well. No better. But SOCOM has some ideas. They had FN scale up the Minimi/M249 light machine gun to take the 7.62mm Nato round. This is the FN Mk. 483, and it fits the bill for a reliable modern 7.62mm machine gun that’s lightweight. SOCOM approved! And it comes in at 8.2 kg (18.3 lbs). Not bad. Beats out the German HK121 as well.

So is that it? Have the Belgians taken the gold? Not quite. There’s still the Israelis to think about. And their Negev NG7 is another scaled-up 5.56mm machine gun. But it’s phenomenally light. Comes in at 7.6 kg. That’s almost PKM weight, and with rails and an adjustable stock to boot. We have our winner. Mazel tov, IWI.

Oh, and if you do have a song celebrating more modern weapons and using those to civilize terrorist scum, drop us an email. Or, write to us at:

c/o This song should have been at Fishbreath’s Wedding
1 Parvusimperator Way
Imalwaysright PA, 16046

1.) I.e. unbelted
2.) You may be more familiar with the American version, the M240.
3.) Not to be confused with the Mk 48 torpedo.