VictorInThePacific

Lest anybody say that the way the Harpoon AI behaves in complex intercept situations is wrong, consider the following real life scenario: An attack on an American CVBG. The only time this really happened was in WW II. Commanding the American fighters, you know that your only base must be protected at all costs. The incoming bombers are escorted by fighters. Do you a) ignore the enemy fighters and go for the bombers, thus losing more of your fighters in the air combat, or b] engage the fighters first and risk losing the carrier? A human will make a choice, which might be good or bad. An AI must choose one option, or there could be some randomness. But there will always be situations where that choice is wrong, and then it is easy to say that the AI is defective. Incidentally, is there any way of changing the emoticon so that my lists don't have emoticons in them?

Thanks, Brad, for doing that. I do understand that setting up that situation takes far more than just a couple of mouse clicks. Obviously I asked you to do it because that's the effect I recall. It looks like it gets more complicated the further you dig. My experience has been that airplanes never get to launch their weapons at ships at maximum range. I always assumed that it had something to do with target motion; that it would depend on which way (and how fast) the target was moving. That is, the GE would simply not let you fire if the target would, at its present course and speed (velocity), run the missile out of fuel. Now it appears that this assumption is incorrect. My experience is also that the Lynx would not in general survive closing to 7 nm from the Nanuchka unless I first diverted the Geckos. My experience is anecdotal. It is not a scientific study. That would require either deconstructing the game code or running a whole bunch of controlled experiments to get a good statistical analysis of the behaviour. The other interesting thing here is: If I understand this last experiment correctly, then the Geckos never fired, either at the Lynx or the Sea Skuas, one of which was shot down by point defence. In real life, this would be explained by the SA-N-4 arc being forward. In RL, wouldn't the ship turn to bring them to bear? Would it have time? Does Harpoon actually restrict SAM fire by RL arcs? Last time I remember this being discussed, I recall that the answer was no. Could it be that the Nanuchka never detected the Lynx or Sea Skua? (unlikely) Or that, perversely, it decided that they weren't a threat? (likely) Sometimes weapons get saved for a rainy day, even when the hailstorm is happening right now. Maybe I will actually run a whole bunch of tests on this, but it would be nice to have a definite answer to the weapons arcs question first.

A further comment on this. I think this maneuver is not only risky, but actually suicidal. As soon as any AI sub gets a sniff of you, every Oscar in the world and his brother is going to unload a bunch of cruise missiles on you. In fact, I consider the situation to be so dangerous that, when a scenario starts, EVERY ship goes to full stop. I figure if an AI sub detects you in the first 30s, then you are toast anyway, but if not, stationary = silent = safe. Well, not EVERY ship. Groups with good ASW assets and the capacity to deal with sub-launched cruise missiles are allowed to proceed at 15 kn. Fighters and ASW planes are sent to escort stragglers to safety. I'm assuming that you have land-based air to deal with AI long-range bombers, and the main threats to the straggler ships are torpedoes and sub-launched cruise missiles. Furthermore, I don't think ship groups should be traveling around without active sensors. If the group has any aircraft available, some of them should be up with active radars, looking for incoming cruise missiles. The aircraft are safe because enemy fighters will be too far away to engage them, and the long-range stuff doesn't have ATA capacity (unless you send LRB to shoot down helicopters with guns???). The AI will not be able to deduce the presence of ships from the presence of escorting aircraft, so this is a "cheat" of sorts, but not really, because I'm talking about ships moving around far from enemy bases anyway, and subs cannot realistically detect aircraft in any case. If singleton ships ARE moving about near enemy airbases, well, they're going to die anyway. In the absence of escorting aircraft, I think that you should use the feature that different units in a ship group can have different sensor states. Low-value units with good radars should have them on to give some warning of sub-launched missile attacks.

Brad, since you've gone to the trouble to test this, would it be an imposition to ask you to test it again, but with the Lynx coming in from BEHIND the Nanuchka?

Kavik Kang: I don't think that the AI has a special hate-on for AWACS. (Again, Brad or Tony can say so for sure.) As I understand it, the AI will attack anything it sees. And AWACS may be the only thing it sees. If an AI fighter is intercepting your AWACS, and it stumbles over your fighters, it will or will not attack your fighter instead. This is probably a choice that definitely goes one way, but I have no idea which way it goes.

Regarding taunts. The action in Braveheart was intended to provoke the English to charge, so that "The Hielanmen, wi their lang swords, [could lay on them] fu sair". It failed. Those nasty, professional English soldiers committed their archers, against which the Scots had no defence. In Harpoon, the defending fighters have some advantages, and the AI can usually easily be provoked to charge.

Joe, from your descriptions, it is clear that you are using very good tactics. Some further comments: AWACS apparently is not useful for locating ground units (eg. SAMs attached to bases). So you really do want an RA-5C or modern equivalent for the mission being discussed. Complicated question: You have a bunch of F/A-18s armed with Harpoons and Sidewinders at the start of the scenario. Do you a) re-arm them with Sparrows right away and risk being hit before they're ready, or put them up as is right away, or c) dither, or d) cheat by playing the scenario twice to learn the future? Your comment on using minimum weaponry is right on. When I say, put all fighters on the front line during the air superiority struggle, I don't really mean that. You need to keep some back as replacements for when the first lot runs out of fuel so as to maintain continuous coverage. What is the proper ratio? That is unclear. 2/3 up front to start with? In any case, the AWACS can't be up there if you don't have adequate fighter protection. So you have to deal with that problem as well. If enemy fighters can freely approach your AWACS, then something is very wrong. I do have a couple of suggestions. First, pickets are not the issue. Detection is from the AWACS, not the pickets. Second, depending on which one it is, an AWACS should have a good air search radar and maybe a good ground search radar. Now, as I understand it, and please correct me if I am wrong, the AS radar does not work at all altitudes. So a plane at low or VL can safely and invisibly drive right up to your AWACS unless it has a good GS radar. Third, maybe you are dealing with real stealth fighters. Fourth, I am assuming that your AWACS is flying as high as possible, otherwise the radar ranges go down. Fifth, good jamming may degrade the AWACS radars. I don't really have any good answers to what I called Phase 2c above. However, by then, your fighters are all expendable, right? You, Tommy, go skirmish those guns! (Sorry, Tony.) Are you really sure you don't have any RA-5C-equivalent in the arsenal? Personally, I might not play such a scenario. If you decide you need to kill all the AI units for your own satisfaction, then maybe you might find yourself removed from command after throwing your planes away unnecessarily. If the mission is to destroy ground units, you may find that you get units with the capacity to actually do the job. Scenario victory conditions are usually given in the orders.

In real life, there would be all possible permutations. What do you need to do? What is the best way to do it? Each situation has its own answer. Same in Harpoon. Same questions. Same answers. Bottom line is, you probably need to do a calculation each time as opposed to just winging it, especially if you want to maximize efficiency. Just like staff officers do in real life.

A rather broad statement that isn't entirely true. For one thing, larger groups will mean a larger radar return. But on the other hand, you might benefit from multiple sensors and weapons. Sorry, Brad, my statement was unclear. "For example, it may be standard doctrine to fly fighters in pairs. But Harpoon does not deal with the scale at which such things matter, so you can group your planes in pairs or not, and it will not affect the outcome in any way, except that the player needs to do more or less work." Here is a better expression of what I was referring to. Consider the Thach Weave. This was initially developed to allow a pair of (WW II) Wildcats to deal with the much more maneuverable Zero. There is no way to simulate this in Harpoon (except by giving the Wildcat a much higher ATA rating than the Zero, which would be incorrect). Consider the (WW II) Flying Tigers. Using P-40s rejected by the RAF as being inadequate, in Burma and China they inflicted heavy casualties on the nominally much superior Japanese fighters. But it was purely a question of tactics, where the strengths of the P-40s (armor and better dive performance) were maximized. Again, there is no way to simulate this in Harpoon. The Thach Weave shows why you might want to operate fighters in pairs. In Harpoon, it doesn't matter. Interesting comment on radar signatures. Now we have to deal with the follow-up question: Is it better in Harpoon to operate two fighters as a group of 2 or 2 groups of 1 w.r.t. detection? I suspect it's too complicated to make a decision. Consider the (WW II) Battle of Midway after the first American airstrike. With only one carrier left, the Japanese finally located American ships. But the Americans were operating in 2 groups, with the Yorktown separated from the other 2 carriers by about 10 miles. The Yorktown was attacked and wrecked, and since the other American ships were elsewhere, the Japanese for a while thought that all the American carriers were sunk. Point is, by spreading your forces out, it is more likely that some of them will be found AND that some of them will not be found. Of course, the total weapons and sensors are the same for 2x1 as 1x2. Why am I describing all this WW II stuff here? Because I just picked up a bunch of WW II books real cheap.

I would like to address the question of attacking static defenses. A lot of other posters have already provided a lot of good ideas, but still, it was rather general, so here’s my detailed “cookbook” approach. I will be using somewhat irreverent language; this amuses me. Please refer to my Thanh Hoa Bridge scenario AAR: http://www.matrixgames.com/forums/tm.asp?m=2104380 This scenario is quite different from modern scenarios, because the Americans have hardly any stand-off weapons (according to the scenario instructions) and they certainly don’t have stealth bombers, but then the Vietnamese also have limited technology. In fact, the Vietnamese in this scenario actually have a pretty weak defence. The American bases are never under any threat, so it is purely a set-piece battle against a static defence. General requirements for dealing with a static defence: 1) Get yourself an RA-5C. Better yet, get half a dozen. This unit has capabilities that other units (including AWACS) just don’t have. It can detect things that do not emit radar signals! Since this capability is so important, I am sure that the modern arsenal includes units that can do it as well. However, I don’t know what those units are. This is a deficiency in Harpoon. Whatever that capability is called, it should be listed in the database unit properties in the same way that weapons ranges are, but it isn’t. Now if you have these recon planes, just follow along the AAR. But if you don’t, new techniques will be required. 2) Be prepared to creep with your airplanes. You may ask, how does a jet plane creep? Well, keep in mind, we are not really concerned with such things. The point is that there will be times when a particular geographical area is free of enemy units, and that area will have a boundary. Beyond that, there could be enemies, in particular, hard-to-locate ones. So you creep your planes forward and hope that the enemies reveal themselves by shooting at you before you can’t escape or you see them before they shoot. Bear in mind that, even if they think they’ve got you dead to rights, you do have an emergency speed that they probably didn’t take into account. As Kavik Kang pointed out, first of all you have to eliminate the enemy fighters. Phase 1: Obtain air superiority Phase 1a: Scramble ALL your fighters in the ATA role. Yes, ALL! Re-arm ALL your available units for the ATA role if necessary. Yes, ALL! DO NOT attack ships or bases in this phase. Phase 1b: Place ALL front-line fighters (with long-range missiles) on the front lines. You may prefer a verb indicating high speed, but keep in mind that we aren’t really concerned with such things here. You want your units to be in specific places. Eventually they will get there. Place your AWACS assets where they will see incoming aircraft before your fighters are threatened. If necessary, use your second-line fighters to deal with whatever gets through the front lines. As an example, a Harrier with Sidewinders may be able to deal with ASMs, but probably not with Flankers. Phase 1c: Taunt the enemy. To learn how to do this properly, watch this video: Incidentally, that is the first hit if you google “taunt youtube”. Otherwise, there is a scene in Braveheart that is also relevant. If you’ve seen Braveheart, you know what I mean. If not, ask yourself what a Highlander wears under his kilt. Ask yourself what you should never do to a werewolf. You will get the picture. Bottom line is, all the enemy fighters are now charging blindly in your general direction. Pick them off with your long-range missiles. Phase 2: Eliminate enemy ground defences Your fighters play little role here. Put all units that matter on the ground attack mission. The front line now consists of attack aircraft and such. Phase 2a: Locate the enemy radars. This includes base radars and attached radars and anything that has a radar. Many of these units don’t turn their radars on unless provoked, so provoke them. Fortunately, all that may be necessary is to paint them with your own radars. Phase 2b: Neutralize enemy radar-guided weapons. You can do this by eliminating either the radars or the weapons. If all the radars are dead, the SAMs can’t shoot. At least, I’m pretty sure that this is true. Use stand-off weapons. Phase 2c: The hard part. You now have to deal with units that are hard to find. But these units will also have short ranges, right? So you may be able to simply ignore them. You don’t need to destroy all these units, do you? You just need them to not hurt you. So clear one narrow path to the objective; that’s all you need. And now I will do what every good teacher does. I will leave the hard part as an “exercise for the reader”.

Yeah, it does depend on the figures used in the database that governs the scenario, obviously. I was referring to the real life or best known figures that the most current database strives to emulate. (Btw, when I look at the GIUK values, I see a value of 8.0 nm for both the Sea Skua and the SA-N-4 Gecko). I was definitely using an older version of the game when I did this exercise (8 nm vs. 7 nm). There is another factor that may also be relevant. I don't have the game available right now, but don't ASM have to be launched at less than maximum range? It seems to me that I have seen this restriction before.

Silent Hunter, thanks for putting all the effort into writing this up. I find the content interesting, and I enjoy your humour.

I wouldn't even attempt this one as the Soviets. How are you going to penetrate the Phoenix umbrella? It's impossible.

CAP may prevent the above-mentioned nightmares.

I would advise against this maneuver. The Soviets generally have air superiority in this scenario. It's OK if you search for a bit and then run back home.

A minor semantic quibble: The Harpoon airplane groups have nothing to do with any real life grouping. For example, it may be standard doctrine to fly fighters in pairs. But Harpoon does not deal with the scale at which such things matter, so you can group your planes in pairs or not, and it will not affect the outcome in any way, except that the player needs to do more or less work. Personally, I consider the following factors (in addition to what the previous posters have mentioned): 1) A given group can only be in one place at a time. In how many places do you need your airplanes to be? 2) A given base can only launch as many airplanes at a time as it has runways. That affects my group size. If the group is larger than the runway number, then the first-launched planes waste 30 s (or more) worth of fuel. 3) I almost never use 1-unit groups for common planes. Why not? Because I don't feel like it. Those are the scientific principles.

Attacking Nanuchkas with Lynx/Sea Skua is a really bad idea (translation: suicide). I discuss this in detail in an old post in the Matrix forums.

I haven't read all 60 replies yet, and this is a couple of weeks old, so I may be repeating someone else's post. Personally, I feel that 4 missiles per good fighter is about right to actually kill it. The Flanker definitely qualifies. That's if the missile has the range to hit. Running away is also an option. 27 nm for the Sky Flash. If nobody has given the actual hit probabilities, I will revisit this. Maybe the Flankers got lucky when they got you? And you got unlucky? Shooting at Keflavik with 32 HARMs should give you a whole bunch of hits, knocking out the radar, but doing little else. That's if there are no SAMs or fighters defending. Or maybe it's possible that you actually wasted an attached radar unit and nothing else? If your ships are innocently driving along and suddenly eat a whole bunch of missiles, it's pretty clear that something saw YOU. Perhaps a sub? And you can assume that enemy has lots of long-range cruise missiles kicking around. The source of the missiles could be far, far away. If your ships are traveling at more than 15 kn, you can be sure that subs will hear you, and you will not hear them. Apparently the AI is not very good at all, but the scenario may be designed to compensate. More later.

Unarmed ship types could be part of the victory conditions: lose this, you lose. Landings can be simulated in the VC as well: these ships need to spend a certain amount of time at this (coastal) location.

Long time ago, but here are my suggestions for GIUK 1: http://www.matrixgames.com/forums/tm.asp?m=2200890 (scroll down to post 14)

Quite the opposite. The Standard ARM which appears in the Thanh Hoa scenario was developed precisely because the Shrike suffered from a short range. So, in fact, rather than deviating from it, Harpoon seeks to model real life. I stand corrected. I was mixing up 2 different time periods in my head. In that scenario, if you allocate your precious and rare Standard missiles parsimoniously (which I did), you can "safely" knock out the SAMs or their radars. Then, by matching up your second-rate weapons against their second-rate weapons, and so forth, you can pretty much retain the range advantage, and thus safety, all the way down the line, and there is only one exception to this process in the scenario.

Actually, I'd have to disagree and say that Harpoon does a pretty decent job of modeling the early Iron Hand or Wild Weasel missions. This was, for the most part, 'reactive' SEAD. In other words, you go out on a mission actually looking to get shot at, and when the enemy exposes himself by shooting, you respond in kind. The SAMs were fairly simple, and so were the ARMs and bombs. Let me clarify my previous statement. As I understand it, what used to happen in real life is a plane would fly over the SAM site, hoping to provoke a SAM launch. This implied that the ground radar (there might be several categories of radar) would have to be on. Then a plane (might be the same one) would launch an ARM at the ground radar. The ARM would typically have shorter range than the SAM, the point being that some plane was always at risk while the SAM site was being attacked. It might be possible for the ARM to be defeated by turning the ground radar off, but that would imply loss of the SAM. So the idea was to provoke the launch of enough SAMs to inhibit the turning off of the ground radar. In the meantime, how does the "bait" plane survive? Apparently, it can outmaneuver the SAM by turning, particularly in the vertical direction (diving), which starts to become problematic if there are too many SAMs in flight. In Harpoon, the first thing you need to do is to get the SAM radar to turn on. This appears to happen automatically if you just paint the ground radar with your own radar. This is a significant deviation between Harpoon and real life, but please don't ask me to write an AI that can play that radar on/off guessing game. At this point, you shoot an ARM at the ground radar. Admittedly, my experience with this type of situation would have to be rather limited. However, I have rather detailed experience with several scenarios. For example, in the Thanh Hoa Bridge scenario, the ARMs outrange the SAMs, so all the SAMs (at least, their radars) can be wiped out without ever risking an airplane. This is a significant deviation between Harpoon and real life. In the USNI 1 (RED) scenario, the Soviet planes are so crappy that there can be situations where the SAMs outrange the ARMs. While it is possible to micromanage some sort of horizontal maneuvering (turning) in Harpoon, there is no possibility for vertical maneuvering. When I wrote "Harpoon does not model such processes very well", that is what I was referring to, and that is the crucial point. However, in the USNI scenario, I was able to provoke a radar on / SAM launch with my "bait" plane in SAM range. My plane escaped by flight (as in fleeing), but not by maneuver. So you can see that you can get some similarity with real life in the game, but if you examine the specific details, they don't match. Not that I am complaining, mind you. As far as I know, Harpoon was never intended to function at the scale of tactical maneuvers by individual airplanes, and it would be meaningless to expect this of the simulation. True, and true. The obvious restriction is that it limits the types of scenarios that you can play effectively, because you can only personally control one group at a time. (The new "pause" function might render this statement invalid.) It is obvious that you can increase the performance of your force by micromanagement. What is interesting is that there are some situations where you can increase the level of realism by micromanagement. For example, airplane maneuvering in Harpoon tends to be one-dimensional, where you move in a straight line from one point to another. There is a certain amount of two-dimensionality in the horizontal direction, where the flight path might direct an airplane around a major obstacle, such as a SAM site. There is a certain amount of two-dimensionality in the vertical direction, where you can change levels. But as I mentioned in my previous post, there are some useful and interesting things you can do if you investigate the three-dimensional space around an individual unit. Of course, if you consider the entire theater of operations, it is certainly three-dimensional on that scale.

If a region is defended by modern SAMs, and all you have is iron bombs, perhaps the correct solution to the problem is to not enter that region. That region could be considered a "no-fly" zone. Patton said that a soldier's duty is not to die for his country - it is to make enemy soldiers die for their country. A brute force suicide charge is not an intelligent solution to a military problem. If that's the only option available in the scenario, I won't play that scenario. I've got better things to do with my time. Having said that, there are advanced techniques available in Harpoon that simulate techniques from the early Iron Hand or Wild Weasel missions. I have experimented with some of these techniques, and reported on them here and there in the Harpoon forums. Bear in mind that Harpoon does not model such processes very well. Harpoon simply does not function at that scale. So any attempt to put these processes in will require a lot of extra micromanagement.

If we're talking about the same scenario: In that scenario, you probably wouldn't really use your forces the way you are instructed. You get a couple of Intruders. But the Intruder has your only airborne radar, so it should be used primarily as a detector! And it can carry spare fuel, so its secondary mission should be as a tanker! And only use it as a bomber when strictly necessary. So if you do what I just suggested, you should have no difficulty finding (and sinking) the surface targets. Finding the sub is another matter. It seems to be that finding subs AT ALL in HCE is a problem, because sonars are pretty poor, except in the most modern units. It turns out in this scenario that some of your ships have been set to crisscross (again and again) the area through which the sub is set to pass. Perhaps it's a noisy sub? But you have to keep your own speed down, and then you still need to get fairly lucky. More on detection: Probably you could say that this is the key to modern air/naval combat. First of all, you have to locate your enemy (any you want him to NOT locate you). You really need to devote a lot of resources to this. In some scenarios, you aren't given proper detection assets. I tend to not play those scenarios ...

I extracted some comments from my Thanh Hoa Bridge report: ______________________________________________________ 2 Hawkeyes and 2 Vigilantes at the front … A Vigilante will be used to locate the Vietnamese ground defences. Vigilantes are fast enough to escape from most threats, and these planes appear to locate ground units very well. We launch an RA-5C, and what a difference that makes! 16 immediate contacts: 4 air, the rest on the ground. About 45 min into the scenario, we have been able to refine the locations of most of the enemy ground defences.The Vigilantes are pushed forward. Within 10 minutes, we start to acquire new targets by infrared. The first contacts are obtained at somewhat less than 15 nm. Loitering the Vigilantes at that distance to increase the scan time does not help. They need to approach to about 10 nm before any detailed information is available about these new enemy units. They are non-radiating, _______________________________________________________ Loss of contact on the enemy fighters … seems to be happening a lot. When this happens, our planes need to run away to avoid bad accidents. _______________________________________________________ Only 6 of our attack aircraft have “stand-off” weapons … Only the KS-19 (AAA) has a range that is equivalent to the rockets on the Corsairs. rockets: 4.3 nm; KS-19 (100 mm): 4.5 nm; KS-12 (85 mm): 3.5 nm; Grail missile: 2.3 nm; ZU-23 guns: 1.4 nm. Iron bombs will only be used against targets that can’t shoot back. Rockets will only be used against targets that can’t shoot back or against the 100 mm AAA. 3 KS-19s will be avoided. They can’t do anything outside of 5 nm range. Two KS-19 batteries are waiting near Thanh Hoa Bridge. KS-19 hit probability: 10% ________________________________________________________ The first section shows how to locate ground defences. Typically, you will need something like an RA-5C, which performs MUCH better in this role than any other aircraft. I am guessing that HCE labels such units as "recon" or perhaps "EW". I don't think that the unit reports tell you which units are good at this. Lacking such units, perhaps the only way you can locate ground units will be when they shoot at you. It takes a long time to properly locate ground units, and you may need to get very close. Put plenty of eyeballs to work. The second section tells you: if you lose contact on a mobile enemy unit, run away! The third section gives some attack details. If you are short of true stand-off weapons, be creative! Line up your units against the enemy units so as to maximize your range advantage (you do have the mobility advantage). In this scenario, although the Vietnamese had well over 100 units, they had a maximum of 2 cases where they even got to try a shot, which would have been less if the LGBs had hit. But 2 chances at 10% isn't very much, and it may well be that the Corsairs spent so little time in the critical 4.5-4.3 nm range bracket that, in fact, the KS-19s never did get a try. In situations like this, it may be crucial to manually keep trying to shoot (hit F1 with your unit selected) as you drive forward.