In a combat, GURPS has us ask defenders this question: “The attack will hit you. What do you do?”
What if, instead, we asked this: “The attack might hit you. What do you do?”
From will to might: that single-word change is arguably a much more realistic representation of the knowledge a defender would have – in many cases, the knowledge that a defender possibly could have – in battle. It suggests interesting implications and tactics, for no extra complexity in play.
I’ve been a vocal (if not quite proselytizing) proponent of the proposal below; I’ll daringly call it “the biggest pro-realism, pro-roleplaying bang you can get from a zero-bookkeeping, zero-extra-rolls, zero-effort rules tweak to GURPS combat”.
Although the article addresses GURPS, it could apply to any game system that runs defense rolls in the same way. Read on!
What do you know?
Active defense rolls are an innovative and nifty feature in GURPS.
The way they work is also a wee odd.
FEND points out how defense rolls are at odds with normal game mechanics in one respect: They’re the only regular instance of a success roll made against a fraction of Skill (Skill/2, and sometimes Skill x 2/3 in 3e). That’s a quirk, however, not a problem, and FEND’s suggestion for full-skill defenses is all theoretical musing, not a “fix”.
Setting that entirely aside, there’s a different oddity with defenses that’s peculiar in two ways: It runs at odds to normal game practices and brings combat realism down a good notch.
Ever wonder why a combatant makes the decision to defend after he learns how the attacker’s to-hit roll fared? Ever wonder why we don’t have him make that decision without knowing the TH result?
Are you wondering why that’s even worth thinking about? If so, consider the below.
What defenders know (or, “Hold on while I read that incoming bullet . . . nope, my name’s not on it”)
Say you’re creating a game that uses the spiffy mechanism of active defense rolls. And as part of the campaign background, you allow powerful psi characters, including amazing precognitives who can even predict whether incoming blows and missiles will hit or miss, and can then choose how to react accordingly. How would you game that incredible precog ability?
Hmm, that one should be easy! When a sword or bullet comes whooshing in, you let the precog know in advance whether it’s on target or not, and thus whether or not she needs to take defensive action. If she senses an incoming miss, she can stand pat and laugh as the danger whizzes by! That’s a simple, intuitive way to game the amazing ability.
Except … in GURPS, everyone already has that ability! Characters certainly don’t have precognition by game-design intent, but effective precognition is the mechanical effect of the attacker reporting “hit” or “miss” to the waiting defender, before the defender needs to make any decision. In GURPS, everyone under fire gets to say “Good, that bullet’s going to miss me” and take no defensive action whatsoever.
The DECIDE defense tweak (or, “Yikes! Bullets! Duck!”)
Here’s a suggestion for an improved way to play: When an attacker announces his attack, have the defender decide his response without knowing the TH result. (The defender has to be specific, as always: say, “I’ll Dodge and Retreat”, not “Yeah, I’ll defend”.) Resolve TH and defense normally.
Or in other words:
Instead of asking defenders “The attack will hit you. What do you do?”, ask “The attack might hit you. What do you do?”
From will to might. That single-word change sums up the entire tweak to character knowledge suggested by DECIDE.
How it plays
Here’s how this works for your defenses in play:
- If you choose to defend and then find that the TH roll was a good one, you made a smart choice that may save your hide. Roll your defense.
- If you choose to defend and the TH roll turns out to be no good, you’re okay. You ducked or otherwise readied a defense against an attack that would have missed anyway, but hey, that’s life. Your head’s in one piece; quit griping.
- If you don’t choose to defend and the TH roll misses, you lucked out – and probably looked pretty impressive (or nuts), standing pat as bullets chew up the wall around you.
- If you don’t choose to defend and the TH roll succeeds, you’re hit. Too bad. Better luck (and duck!) next time.
Note that no new rolls or even rules are involved here. We’re simply talking about intuitive effects of player decisions.
How to limit character knowledge
How do you keep a defender from knowing the attack’s TH roll? That depends largely on how you handle combat rolls in the first place.
If you roll TH openly, simply have the defender state his action before you roll TH. The attacker states what he’s doing, the defender states his response, and then you roll TH (and if necessary, defense).
If you roll TH “secretly” (which is pretty common behavior for GM rolls), you can roll TH before or after the defender states his action, as you like; it doesn’t matter. All that matters is that you ask the defender “What do you do?” before revealing the TH result.
Ranged or melee?
You’ll probably agree that this all makes sense when defending against gunfire; a normal character couldn’t possibly know that an incoming bullet will miss and thus require no defensive action. DECIDE is a great rules change for that situation.
But how about incoming swords and spears? Could characters reasonably gauge these attacks accurately enough to know when a defense won’t be necessary?
The melee case is debatable. Here are three options for where to introduce the defense tweak:
Option 1: Firearms only, all-or-nothing
Use the defense tweak only for bullets, beam weapons, and the like, where preemptively ducking away from (or blocking, if possible) the perceived path of fire is the only logical defense. That removes GURPS‘ big existing oddity with precognitive bullet-dodging.
For other attacks, where oddities are less pronounced, don’t bother changing anything; let defenders know the TH result and then make defense decisions, per normal GURPS.
Option 2: All attacks, all-or-nothing
Use the defense tweak for any attacks – bullets, beams, arrows, knives, punches. As with bullets above, you defend now or not at all. If you don’t choose to defend and the TH roll succeeds, then you’re hit. Too bad. Try ducking next time.
Option 3: All attacks, immediate vs last-second defense
This option applies the defense tweak to all attacks, but much less harshly than does Option 2.
You have a choice of 1) DECIDE’s “immediate defense” described above, in which you commit to defense as soon as an attack is launched; or 2) GURPS‘ “last-second defense” (if I may name it such), in which you calmly confirm the need to defend before committing to defending.
Give all immediate defenses a +1 bonus. That’s a fair bonus for immediately making the commitment to defense, even when your foe’s mace might miss by an inch anyway.
If you don’t make an immediate defense … wait, you’re not hit just yet. You delayed for a split second to gauge the attack’s exact path (i.e., to check TH) – and have confirmed that a mace-on-head impact is indeed imminent. Time for a last-second defense!
A last-second defense takes a penalty. This would be a reasonable place to introduce a penalty that varies with the speed of attack: say, 0 for a melee attack, -2 for a thrown weapon, -4 for a missile like an arrow, -8 for a bullet, or -16 for a hypervelocity round. (That’s pretty well impossible for bullets and faster – because it represents actually dodging or blocking the bullet itself, not jumping out of the muzzle’s path.)
The Dodge penalty versus a laser beam would be effectively infinite. If you didn’t dodge the perceived path of fire before the trigger is pulled, there’s no way you could even see, let alone dodge, the beam after it leaves the gun. (It’s not a problem in a cinematic game, where “blaster” beams travel at a speed that makes arrows look fast.)
DECIDE’s core idea is one that many players will agree with, but how to implement it is open to any number of ideas. Play around.
You could, for example, tweak both the the immediate defense and last-second defense bonuses/penalties for Option 3: award +1 on Active Defense for immediate defense vs any attack, melee or ranged; and no mod for last-second defense vs any attack, per standard GURPS. Just those two choices, simple as can be. Nothing comes up in play at all, except that +1 bonus for defenders who commit to immediate defense.
DECIDE leads to interesting tactical considerations. It forces defenders to make decisions about doling out limited defenses, or spending defenses that have consequences (like disrupting Aim), with Option 3 further dividing the defense strategies of skilled and unskilled combatants. (Note too that “immediate defense” meshes well with defensive actions such as All-Out Defense, Retreat, Feverish Defense, and Defensive Attack, while “last-second defense” fits nicely with a more aggressive stance.)
See Benefits of DECIDE in the Appendix below for plenty of notes on tactical implications of DECIDE.
This article isn’t saying that current defense rules are nasty, evil, or even “broken”. There’s no game-wrecking flaw in the workings of defenses as written; the worst I can say about it is that it’s an odd handling of character knowledge. (I didn’t notice anything funny about it for years, so it can’t be too big a problem!)
I think the DECIDE tweak is easy, intuitive, more realistic, and more fun to boot; “the biggest pro-realism, pro-roleplaying bang you can get from a zero-bookkeeping, zero-extra-rolls, zero-effort rules tweak to GURPS combat”, to make a bold claim. But naw, it won’t change your life, make you taller, or lure actual girls to your gaming table. And in many combat exchanges, it won’t even make a mechanical difference in outcomes. As always, give it a try, and adopt or refuse as you like.
While the above is all you need to play the tweak, below is more for the interested to chew on, such as further info on merits of the tweak, notes on objections raised, and a closer look at how much it even matters in play.
First, a few notes on the current DECIDE article compared with an earlier version:
Notes on the article revision
The first version, “Revised Defense Flow in GURPS“, was an unexpected hit in terms of visitors and (mostly very positive!) comments, but I wrote it poorly. I focused on the order of gameplay – specifically, rolling TH before stating defenses. Yet if one thing’s clear from discussions on the topic and play experience, it’s that focusing on the order of mechanics invariably leads to misunderstandings.
That’s my fault for aiming at the wrong target. How to order gameplay is really a secondary matter of implementing the suggested change. The core issue itself is character knowledge – specifically, whether a defender should know in advance that an attack will hit (or miss) when the blades, arrows, and bullets start to fly. Rewritten with that focus in mind, DECIDE should strike even more readers as sensible.
The topic comes up a lot in online discussions. This article brings together everything I have to say on the matter, drawing on previous online discussions and writings here, including:
http://forums.sjgames.com/showthread.php?t=29004 (online forum discussion)
http://www.gamesdiner.com/GULLIVER/B5combat.htm#DeclaringDefenses (very old, 3e-era initial write-up)
https://www.gamesdiner.com/topics/decide/ (all DECIDE-related content on this site)
1. I moved much of the text to the Appendix; it’s supporting material, not key content.
2. I changed the term for “defending only after confirming TH” from the initial “panic defense” to “last-second defense”. The former was a poor choice of words: You could argue that the “immediate defense” (initiating defense without checking TH) is one taken in reflexive “panic”, and the delayed “last-second” defense is one chosen deliberately and calmly!
3. As mentioned in the original version, the tweak creates an ability to force a defense using a Feint. When you announce your “attack” (really a Feint), your foe may announce a defense, and thus needlessly spend the defense to avoid an “attack” that doesn’t come.
But while that creates interesting opportunities when using the Feint + Attack option for All-Out Attack in GURPS 3e, changes to Feints in 4e make gaming this more tricky. I’ve removed this topic from the article and will set it aside for now, probably until after I see whether the upcoming Martial Arts suggests something. For now, it’s easiest to assume that Feints bring up no special consideration: just as the “attack” doesn’t actually happen, the defender’s stated defense versus it doesn’t actually happen, and the defense isn’t “used up”.
“Objection!”: Readers respond
Does GURPS really hand defenders an unnatural precognitive ability? Some GURPS players do object to this portrayal, or to other aspects of the tweak:
“But a trained warrior should be able to predict whether incoming blows will hit or not.”
Sure, to some degree – but GURPS makes no distinction concerning “trained”. A wheezing librarian who’s battled nothing beyond dust bunnies is granted the same automatic, unfailing predictive sense as a champion gladiator.
Experience is a moot point anyway when bullets and lasers start flying; it makes no sense for any normal mortal to say “That bullet’s going to miss me; I don’t need to defend”.
“It doesn’t matter anyway whether the defender states he dodges.”
“Dodges are free and unlimited. We can assume the defender always chose to dodge.”
No, GURPS says that it does matter, in concrete ways. The game is very specific on the consequences of active defenses: If you’re taking an Aim, Concentrate, or similar time-consuming action, dodging instead of staying put can disrupt your action.
Besides, there are more active defenses than Dodge! The above consequences of defending apply equally to Parry and Block – and unlike Dodge, your choice of one of these “burns up” a limited use. (GURPS even offers published options that place Dodge on the same “limited use” schedule, with each Dodge past the first becoming more difficult. Under that sort of option, “reading” whether you do or don’t need to dodge an attack matters a lot.)
“There are other ways to interpret a miss.”
“We don’t have to say that the defender foresaw the miss and knew he didn’t have to defend. We could say that the miss happened because the target defended in some unspecified way.”
Creatively interpreting outcomes is a good thing that all gamers should do. But take care with interpretations that conflict with specific game rule effects. Decreeing that an attack’s “miss” was actually a successful Dodge by the defender conflicts with rules if the target was Aiming or Concentrating (why didn’t the “Dodge” disrupt those?), if the target had taken an AOA (why was he able to “Dodge” at all?), and so on.
It’s much simpler to be clear on rules items: An attack is either on target or isn’t, a target either defends or doesn’t, and so on!
“The tweak means extra complexity/bookkeeping/rolls.”
No to all! The DECIDE tweak adds no complexity, bookkeeping, or rolls. Nor does it reduce those. It’s not about those mechanical things, and doesn’t affect them. It’s about what the character knows when he decides whether and how to defend. That’s it.
“But asking the defender to state his action is an extra step.”
GURPS already requires defenders to do this – sometimes (that is, whenever TH is a success). Yes, the tweak asks that defenders do so any time they want to maximize defense (by committing to an immediate defense). If that doesn’t sound good, use the tweak only versus gunfire, and only when it really matters (such as when the defender’s Dodge might disrupt some action). Or don’t use it at all.
But let me note this: Players stating how they respond to immediate life-and-death danger is the fun part of combat! It’s something I want to do when my PC is about to be skewered. If reacting to an incoming spear with “I Dodge!” is a bothersome chore to be avoided, perhaps the players would prefer some game system with passive defense rules that require zero involvement by the defending player. (That would be most game systems, if I’m not mistaken!)
On realism (or, does it make sense to dodge gunfire at all?)
A bit of a tangent: GURPS allows characters to dodge gunfire. Me, I’m fine with that.
Real people can’t evaluate and dodge incoming bullets, of course, but the game has a rationale: You’re not ducking actual bullets, you’re dodging away from the direction of the gun’s muzzle, the perceived path of fire. (And while we’re at it, the game is clear that this Dodge is an intentional action, not an effect of your random movement. For all your bobbing and weaving, GURPS says you get no Dodge vs gunfire from the rear, after your All-Out Attack, or in any other situation where you wouldn’t otherwise get to make an active, intentional Dodge. A GURPS Dodge vs gunfire is, by the rules, something you do purposefully.)
Arguably, ducking away from an invisible line of fire should be much more difficult than avoiding a solid weapon, especially versus long-range gunfire. But the issue of difficulty aside, it is reasonable to allow some sort of Dodge vs the perceived path of gunfire. If GURPS wants to cinematically handle this with a full Dodge at no penalty, I’ll go along with it, especially considering how quickly a GURPS bullet can end a PC’s career.
My minor beef, though, is that the game turns this all upside down through unnatural character knowledge! Instead of forcing defenders to preemptively jump away from a gun muzzle’s direction, GURPS expressly lets them wait for the gun to fire, then unerringly gauge the path of the supersonic bullets, then decide whether and how to defend, and then take defensive action.
Wow, that’s busy!
The above oddity holds true, if less glaringly so, for other attacks. Do you have the luxury of knowing whether an incoming arrow or spear will hit or miss, or do you reasonably toss up your shield as soon as you see the missile coming your way? And whether or not a trained knight can automatically gauge the accuracy of an approaching sword blow, should everyone be able to do the same? How about you? Would you, and could you, predict whether that sword blow will strike true, or would you move to protect yourself with a parry as soon as the blade starts sweeping down at you?
It’s debatable. Fortunately, you can play it either way.
On character knowledge and roleplaying
In addition to realism issues, the knowledge GURPS gives defenders is a departure from the normal flow of game action. What’s the basic “unit” of RPG play? I’d call it this little exchange or interaction:
1) Something – some event, stimulus, etc. – happens.
2) Players weigh that something, and, without certainty of the future outcome, state how they respond.
3) Based on that interaction, the GM, players, and game rules determine the outcome.
1) GM: “You hear footsteps approach. It sounds like a guard.”
2) Players: “He might see us trying to break in! We’ll need to take cover. First, we’ll . . .”
3) Everyone plays on to resolve what happens.
Defense rolls, by curious fiat, twist that 1-2-3 order to 1-3-2. That is, an attack is launched (that’s the 1), the GM determines and lets the characters know the “tentative” future outcome (that’s 3), after which the defending player can, based on that knowledge of the future, choose his response (step 2) and – possibly – change the future (go back to 3, get actual outcome). If the rest of the game were played that way, it’d look like this:
1) GM: “You hear footsteps approach. It sounds like a guard.”
3) Players: “Will he see us?” GM: “Actually, no, his Perception roll will fail.”
2) Players: “Okay, we’ll keep working on the break-in.”
To me, running defenses like this emphasizes dice-rolling over the most fun element of table-top gaming: character interaction and decision. Attackers do have to make decisions (that’s a good thing!), yet defenders don’t necessarily have to; the game passes them by, jumps right to the mechanical stuff, and only asks the defender to make a decision if the mechanics call for it. Yet it’s the defender who’s facing a life-or-death instant!
There’s a halberd screaming down toward your character’s head. He’s a split-second away from death. It’s as ‘cliffhanger’ a moment as you’ll ever find. If that’s not a time for you to decide to do something, and decide now, then when the heck is?
There are plenty!
1. The tweak models how you’d defend in real life. If you know someone’s about to shoot at you, you take action now. If you wait to watch the bullets fly first, you’re too late.
It’s still a realistic consideration – if less imperative – when applied to non-gunfire attacks.
2. The tweak hews to the way everything else in the game already works: characters respond to threats or events without knowing in advance what the outcomes of those events would be. Why make active defenses the sole exception to this?
3. The tweak emphasizes roleplaying, not mechanics. “Here’s what happens; what do you do?” is the heart of the game, the source of excitement. An arrow is whizzing in from high above; do you jump for cover or not? It should be an important decision to be made now, not something to be set aside while dice are checked.
Similarly, melee gains closer interaction between two combatants (it becomes more of a “dance”, if you like that common metaphor). Each threat by one fighter must be met by a decision by the other fighter, whether that decision is a commitment to defensive action, a commitment to not defend, or a split-second of calm evaluation before deciding (the last-second defense). No more “Meh, I’ll do something if I have to” as the universal defender decision!
4. The defense tweak emphasizes tactics. Now that a fighter no longer knows an arrow’s future in advance, he’ll have to make tough decisions. Should he hold his shield in reserve against what’s probably a wild shot, or play it safe and fully prepare to Block? Does the fighter defend and risk breaking his Aim or Concentrate action, or maintain the action and hope his attackers are poor shots?
5. The defense tweak creates an automatic mechanical effect of being able to focus on a single foe, as opposed to having to potentially respond to many. When you’ve got a single foe throwing one attack at a time, you’ve got no reason to “hold back” a defense; you can comfortably commit to it every time to claim that +1 bonus. In short: Defending is easier when you’ve got only one foe to worry about and you know exactly where every attack is coming from.
6. The last-second defense option allows experienced fighters to watch and gauge the accuracy of incoming attacks. Anyone can choose to do this, actually, and then make a last-second defense if the attack proves accurate. But an untrained fighter, with his lousy defense scores, can scarcely afford any defense less than optimal (even when the difference is small in the case of melee attacks); his waiting seriously reduces his already-slim chances of saving his hide. He’s best off reacting instantly to any attack!
The veteran fighter, on the other hand, can afford to pass up the benefit of immediate defense. She coolly gauges the attacks and doles out her defenses only when they’re needed. It makes for an interesting difference in combat styles.
7. The last-second defense option also allows a distinction between dodging the “path” of gunfire, which GURPS lets everyone attempt, and dodging the bullet itself, for which there’s currently no option. If you’ve got a Matrix-style game with super defense scores, characters can now try to actually dodge bullets!
8. Going back to this article’s opening, the DECIDE tweak creates the perfect way to game real combat precognition. Just apply GURPS‘ current defense rules to precogs only. The result: Everyone in the party gets told “The orcs start firing arrows. What do you do?” But the precog is told “Two arrows are coming for you … but … GM checks TH … both will miss you by a hair. What do you do?” And the precog can just stand there looking smug, if she likes.
9. The benefits of all this are free! A time-tested GM’s first reaction to all the above will rightly be, “Ah, but what’s the catch? What new combat steps, extra rolls, added bookkeeping, and worse will we have to undertake?”
NONE! There are literally no extra rolls or bookkeeping. No funny extra phases or steps. Just take the same “Here’s what’s happening; what do you do?” play that you use everywhere in the game, and apply it to defenses too, giving defenders no future knowledge they shouldn’t have. You still carry out normal TH/defense exchanges on the spot, just like now, with nothing extra to record for later.
Another optional benefit: Player-focused “roll reversal”
If you like, DECIDE can reward your play with more dicing by the player, and less dicing by the GM. Here’s how:
Under normal GURPS play, for any TH/defense exchange, you only go to the final step (damage) if two conditions are met: TH succeeds and defense fails. If either of these is unmet, the exchange ends.
When a PC attacks, the player always gets to roll TH, which is fun for the player. Afterward, the GM only rolls defense for the defending NPC if necessary, a time-saver which is usually welcomed by busy GMs. This is all good.
But now we move to the NPC’s counter-attack. Under current rules, the GM always gets to (or has to, as he may feel) roll TH when the NPC attacks. The player waits passively; he gets to state an action afterward and make a defense roll only if it’s necessary. Yawn.
Hmm. Now let’s bring in the DECIDE tweak. It’s the NPC’s turn to attack. Before any TH is checked, the PC announces he’ll play it safe and Dodge. (Already, the player is getting involved! Yay!) The scene is set; it’s time to resolve it. But… instead of rolling TH, the GM hands the dice to the defending player. That’s right. Both the NPC’s attack and the PC’s defense have been announced, and either a failed TH or a successful defense will end the exchange. That means the order of the dice rolling no longer matters. So let the player roll defense first. If he succeeds, the exchange is over; move on. Only if he fails does the GM need to check the NPC’s TH.
If that observation doesn’t spark a small “wow”, please read it again. Nothing changes in the number of rolls needed: one roll is required, a second only if necessary. Nothing changes in the outcome probabilities (other than the first roll now allowing a chance of defense criticals instead of attack criticals). Yet you’ve guaranteed that the player will always make a roll during both the PC’s attack turn and the NPC’s attack turn, while during either attack turn, the GM will make a roll only if necessary.
This is purely an option, for use only where it adds to the fun. If you care about side effects of the NPC’s TH roll like critical hits and misses, or hitting the wrong target on a crowded field, then you definitely want to roll the NPC’s TH first. But if you’re only concerned with simple hit/miss by the NPC, such as when scrub foes surround the party, the above “roll reversal” keeps the action – and the dice! – focused squarely on the PCs.
Now, what player won’t like that?
Does it all matter?
As “correct” as the defense tweak is, it often won’t make a mechanical difference in play. For starters, many adopters will likely use it in gunfire situations only (Option 1). And even if you think there should be some difference between defending against a melee attack right away vs waiting to gauge its acccuracy, Option 3’s difference in defense rolls is a meager +1. Is that meaningful enough to bother with? It’s up to you.
Likewise, in the common situation of one-on-one melee in which your foe attacks you only once per turn, there’s typically no tactical consideration of whether to “use up” your Parry or Block versus the first incoming blow. GURPS gives you that defense “for free”; there’s no reason not to use it. So in terms of mechanical consquences, in this particular instance, it doesn’t really matter whether we demand that the defender commit to parrying, or whether we just assume that he does so without asking.
Similar things hold true even against gunfire. If the defender isn’t Aiming, Concentrating, or otherwise suffering any consequence of unlimited free Dodges, then no, it doesn’t matter mechanically whether he says “I Dodge!” up front, or whether we just assume he automatically did so.
And so on. In short: the suggestion affects mechanical actions and outcomes in some situations, but not all.
That said, for those situations where it does make a difference, it can be an important one. And there are many side benefits (see above).
Maybe the most important benefit is this: In those combats where the tweak doesn’t significantly change mechanical outcomes, its take on character knowledge is still … well, correct. And without question, doing things “correctly” from the start prevents unforeseen confusion, patches, and game-stopping moments of “Why do things work this way?” down the road.
But all that said, never forget: No matter how technically or realistically “correct” any rule is, where your game is concerned, your response of “No thanks, we still prefer the other way” will always trump “correct”, a thousand times over!
The article is a revision of the old article “Revised Defense Flow in GURPS“. If you see reader comments that don’t seem to fit the article, they probably precede this revision’s posting and refer to the original version. See the Appendix for more notes on the revision.
The DECIDE name comes with assistance from B. Finney, who made many fine (and sometimes wonderfully convoluted : ) suggestions. While the article vigorously fought all attempts at the perfect acronym, we finally hammered out the not-so-bad DECIDE.
v1.5 (2011/02/08): Hmm, I haven’t been putting version numbers on DECIDE; let’s call this 1.5.
The version updates the article with a significant change. Under Option 3, I’m inserting a change first discussed in DECIDE implementation notes: a change from the -1 “last-second defense” penalty vs melee weapons to a 0; and a blanket +1 to all “immediate defense” rolls that don’t wait to check TH.
This is arguably how the option should have been designed from the start. It lets fighters make normal GURPS defenses (“first check TH”) vs melee weapons at no bonus or penalty, and awards an attractive +1 penalty vs any attack when the fighter quickly commits to defending. The net effect: DECIDE becomes nicely “invisible” in melee combat, staying out of play unless a fighter opts for its effects. That’s the sort of option I like!