Page 1 of 1

Wait timer does not count down with aim on?

Posted: Sun Aug 12, 2012 8:06 am
by jonwong
The simple bug report... Aim at some place. Pull out timer (say 1 second). Duck at 1 second mark. The entire action is: Aim at some place for 1 second, then duck.

The complete bug report... Stand at corner for 1 second, then move back from corner (out of line of fire). Then move to nearby cover. Aim for 1 second, then duck.

The command list: Wait 1 second. Move to waypoint 1. Move to waypoint 2 (behind cover). Aim. Wait 1 second. Duck.

The last duck command does not execute. Reason being that the last 1-second wait does not count down.

The reason I wanna aim for 1 second is to fire off a shot before ducking finally.

Re: Wait timer does not count down with aim on?

Posted: Sun Aug 12, 2012 2:39 pm
by jonwong
Oh, I see the problem now.

The timer only counts down when an action is done.

The act of aiming (not the command of aiming) counts as an action. The timer will freeze until the soldier completes his aim and fires, and then the timer will continue counting down.

This removes the ability to give such a command: aim for at most 1 second, then get the heck out of there.

Not that this matters, because if the simulation shows that the aim cannot be completed before the soldier gets shot, then I usually give some other command (like feinting).

Still, it'll be nice to have a command like aim for at most 1 second, then Continue on Sight.

I'm talking about the actual act of aiming, not the command to aim.

In short, it'll be great that the timer continues counting down even when the soldier is performing the act of aiming.

It's been a year since I left this game. I'm surprised no one has spotted this? Hmm. Maybe I should take my feinting skills online and see what happens. :) Still enjoying the campaign now, and still pestering Mode 7 to let me help make an XCom game outta this.

Re: Wait timer does not count down with aim on?

Posted: Sun Aug 12, 2012 7:41 pm
by Scorpion0x17
Yeah, it's been spotted before. Not sure it's ever been reported as a bug though.

But, that's probably because it's not a bug.

It's engaging with the enemy that stops the delay count-down, not the aim command.

So, for example, if you move to a way-point, aim, delay for one second, then duck, and no enemy comes into line-of-sight with that unit, during the delay, then he will indeed duck 1 second after aiming.

Oh, and I agree with you that it would be nice to be able to issue 'aim for one second and then duck, no matter what' type commands - but then you'd also need to be able to say 'aim for one second and then duck, unless you see an enemy', and so on - which would probably greatly increase the complexity of the user-interface.

Need variable length turns, not more complex user interface

Posted: Mon Aug 13, 2012 2:26 am
by jonwong
Actually, the interface doesn't need to become complex. The game mechanics does, though.

The problem is the current 5-second turns. That makes a lot of real world situations impossible in Frozen Synapse (which is also why it probaby can't yet be fleshed out into an X-COM game).

For TL;DR: Frozen Synapse needs to add a Time to React mechanism on top of it's fabulous Time to Kill mechanism.

Also for TL;DR: Frozen Synapse is considered rather luck-based by many gamers (and my chess-playing friends) because it does come down to being lucky in guessing your opponents' moves (also see this Steam forum comment). It is nowhere near true tactics, which is why traditional turn-based tactics (X-COM, Heroes of Might and Magic, Real-time Tactics games, even Real-time strategy games) are still gonna sell to more gamers than Frozen Synapse can ever hope to.

Don't get me wrong, artificial rules do work in tactical games. Think chess.

Now, let's consider why the 5-second turns make for very artificial situations.

You aim out of a window for 1 second, because that's how long it takes for an enemy to come into your field of fire. You then duck and aim at the doorway, because you anticipate he might give the window a miss altogether.

The turn could play out thus: You shoot the enemy from the window. You then duck and aim at doorway for 3.5 seconds because you were primed to do so, even though you know there aren't anymore enemies in your vicinity.

Sounds fake? That's because reasonable human reaction time to self commands is at most 0.5 seconds (human reaction time to external stimulus is 0.2 seconds).

Now, what would make it more realistic? Making the turns variable in length. But wouldn't that make it unfair? Not if we consider the reaction times, ie time to alter self-commands, according to current action. Let's call this concept Time to React.

For eg, if a soldier's current action is aiming out a window, his Time to React could be 0.2 seconds. That means I'll get to issue orders for him again in 0.2 seconds. Comparatively, if the current action is sprinting, the Time to React is 2 seconds (1 second to realize an enemy just popped up, 1 second to stop the run). For a more complex eg, a current action of walking and aiming could have a Time to React of 0.5 seconds, longer than that for aiming while stationary.

Now, you may think that taking a turn every 0.2 seconds is tedious (suppose all your soldiers are just aiming stationary). There could be "many turns", even hundreds, right? The new mechanism is only "tedious" for the server; it might need more harddisk space to store saves and replays (assuming it is aggressively optimized for storing 5-second turns). In fact, the ability to issue new commands at frequencies based on realistic Time to React will certainly make gameplay terribly fast. Why so? You remove the artificial planning paralysis for 5-second turns. Moreover, the game plays more aggressively, rather than force players to camp at windows paralyzed by fear that "too much can change in 5 seconds", and by the inability to react to most changes in that long 5-second block.

All actions could incur a Time to React cost, even standing up or crouching down (say 0.8 seconds?). And you'll soon see that the entire game becomes intuitive. No longer are players paralyzed by having to plan for large 5-second blocks. Instead, players find themselves thinking about real world Time to React when considering which commands to commit. Do you sprint forward at the risk of an enemy popping up from the side and shooting you within 2 seconds? Do you wanna spend 1 second aiming out a window while an enemy slips through from behind?

IIRC, you like single player better. Given the above mechanism, you'll be able to have 4 soldiers beat 10-20 simply because you can give them very tight leap-frog commands (for cover angles) (no longer the large-grained 5 seconds or bust turns). Moreover, we'll be able to make an unprecedented X-COM game with the above, maybe even better than the upcoming new one from Firaxis.

I'm pestering Mode 7 to work with me on an X-COM game, offering efforts in game design, coding, etc. Failing that, I'll probably have to do it myself. Don't get me wrong, Frozen Synapse is unprecedented in introducing the concept of Time to Kill (engagement outcomes). However, the next quantum leap is to add Time to React (command issuance frequency). The first game that does that concept will probably get a perfect 10 from Edge Magazine reviews (the harshest reviewer).

Speaking of which, why isn't there an official list of Time to Kill numbers? This is a chess game after all, and players want to play tactics, not play guess the game's hidden play mechanisms.

Bottom line for Frozen Synapse? Should we still play it? Is it broken? Frozen Synapse is the forefront in psychology games (think Poker), but not at all for single player games. Why is that? Single player gamers get a kick out of fighting overwhelming odds (4 to 20?), and doing that requires more real world and more intuitive control of soldiers for tactics.

Why is Frozen Synapse not the forefront in chess games? Because chess has better Time to React factor: you can react to your opponent's every move. For that reason alone, Frozen Synapse will never overtake chess as a true test of logic skills. It is at most a test of psychological ability to second-guess.

However, Frozen Synapse does let players react to opponents' 5-second plans after the fact. And that's the key to winning matches here: Put actions that require possible change-of-course at the end of your 5-second blocks (so that you can change orders to react to opponent's moves next turn). For that reason, Frozen Synapse is 40% chess, 50% psychology, 10% real urban warefare tactics. Considering that chess is 90% chess, 10% psychology (for competent players), you'll see why Frozen Synapse is more fun (it has urban warfare tactics!).

If Frozen Synapse adds the Time to React mechanism, it could well become the de facto urban tactics trainer/simulator world wide! Not to mention the immediate boost to world tournament worthy game status. And soon, you'll see resumes putting "Champion in Frozen Synapse Tournament" right beside "Grandmaster in Chess"!

Re: Wait timer does not count down with aim on?

Posted: Mon Aug 13, 2012 3:56 am
by Scorpion0x17
So, what you're suggesting is:

I submit one 0.2second's worth of command to the server, and then wait two days for my opponent to submit his 0.2seconds worth of command?

No thanks.

Maybe I explained it wrong?

Posted: Mon Aug 13, 2012 7:33 am
by jonwong
Scorpion0x17 wrote:I submit one 0.2second's worth of command to the server, and then wait two days for my opponent to submit his 0.2seconds worth of command? No thanks.
For TL;DR. Command Issuance Cost game mechanics allow players to:
  • Opt for go-see-go approach (Incur 0.5 seconds for each command issued), OR
  • Opt for go-go-go approach (saves a lot of time)
The mechanic is best seen in real life. You can spend 0.5 seconds to issue a chain of 3 commands (say "take keys, open door, enter room"). Or you can spend 1.5 seconds to issue the 3 commands separately (say "take keys, watch out for possible threats, open door, think what's next, enter room"). Each "turn commit" may even be much much more than the current 5-second blocks due to such time-savings (the 0.5 second for each command). In a typical board game analysis of urban tactics, as much as 20 seconds of "chained commands" are issued because we need to "flash bang, clear room, immobilize hostages, cover exits" all without pause.

Frozen Synapse currently allows 1 option: go-go-go. The only "see" part is at the end of the 5-second block, which was why I offered the tip of "putting actions that possibly involve change-of-course at end of 5-second block".

For non-TL;DR...

Most game mechanics seem non viable until someone implements it. Before Steambirds and Frozen Synapse, everybody said "no" to a hybrid of real-time tactics and turn-based play.

What you don't realize is that it's actually 2 minutes of gameplay (planning) per turn, not 0.2 seconds. Also, turns do not end when the other player still cannot issue a command.

It's also my fault I didn't fully flesh out the game mechanics involved (thought Mode 7 would bring me in to discuss game design, that'll be faster).

First concept I missed: Turn Commit Condition: Can only end turns on other player's ability to issue commands
Second concept I missed: Command Issuance Cost: Each issuance of commands (1 or more) incurs a 0.5-second freeze (the soldier has to copy the command, right?)

The second concept explained in more detail below. For now, let's consider the "No thanks to 0.2 seconds of play per turn per day" concern.

Gameplay will be Faster Than Frozen Synapse Currently

No planning paralysis. Let's illustrate. Say we have a turtle vs an aggressive player.

Aggressive Player has set all his soldiers to a single sprint command, all taking 2 seconds to complete. Turtle Player will be forced to play all 3 turns of Aim commands in 1 session. Why 3? Each Aim command really costs 0.2 + 0.5 seconds (Command Issuance Cost is 0.5 seconds). In short, gameplay session can only end when the other player can issue commands. So Turtle Player is forced to commit all 3 turns of Aim commands in a single session.

And then you may think that Turtle Player can take 3 hours for each turn in aiming, taking 9 hours total! Not true, because there is no planning paralysis. Of the 3 turns, Turtle Player still sees Aggressive Player's soldiers sprinting in the same direction (each 2-second sprint command is a straight line). Does Turtle Player need to re-plan for all 3 turns? Obviously not! We all know where the Aggressive Player's soldiers are gonna end up in 2 seconds!

Gameplay rewards Turtle and Aggressive Accordingly

Now, let's consider how turtle and aggressive plays are BOTH rewarded accordingly. There's a season for all things, a time to wait, to be cautious, and to rush in. That's how things are in real life, right? The game should not penalize turtle play or aggressive play.

The Command Issuance Cost concept also ensures that games won't turn into a Turtle Or Nothing or a Aggro Or Nothing imbalance.

Let's take the act Fire (firing a gun) to be 0.5 second. Let's have 2 opposing soldiers sprinting out of own corners (2 seconds), aim (0.2 second) and fire (0.5 seconds) at each other.

Aggressive Player takes a go-go-go approach, issues a single command session for all 3 actions, incurring a total cost of 0.5 (command issuance cost) + 2.0 + 0.2 + 0.5 = 3.2 seconds. Turtle Player issues a single sprint (0.5 + 2 = 2.5 seconds) and a separate aim and fire (0.5 + 0.2 + 0.5 = 1.2 seconds), for a total of 3.7 seconds.

By the 2.5 second mark, both soldiers have cleared their corners. Aggressive Player cannot issue any command because his soldier's "chain of commands" is 3.2 seconds. Turtle Player's "chain of commands" is only 2.5 seconds, so he can react to the situation (issue a new session of commands).

How does this benefit Aggressive Player? Turtle Player attempts to aim and fire. These commands cost 0.5 + 0.2 + 0.5 = 1.2 seconds, which fires off the shot at the 3.7-second mark. However, he is shot by Aggressive Player at the 3.2-second mark! Score 1 to Aggressive Player!

How does this benefit Turtle Player? He knows he was slow (due to go-see-go approach). He attempts to duck back into cover, which costs 0.5 + 0.5 (to clear short distance to cover) = 1 second. He successfully evaded Aggressive Player's fire 0.2 seconds ahead of time! Score 1 to Turtle Player for wasting his opponent's bullets! (I know Frozen Synapse doesn't have concept of ammo, yet.)

(Sorry, you'll have to draw out the separate time lines to dig the maths above. But these calculations are to be handled by the game/computer, not by the player.)

But wait, what's the real benefit to Turtle Player for taking no shots at all? First, preservation of lives. Second, ability to respond to more situations. Third, and most important, distract Aggressive Player. In fact, I do such plays all the time in Frozen Synapse. It's called feinting, and stapling the opponent in place while another soldier shoots from another location. Without the feint, 1 or both of my soldiers would be dead. I usually feint with 2 soldiers to take out 1 enemy behind cover, all from head-on approach! No need to waste time flanking!

But hold on! Which soldier plans to "round a corner and squeeze off a shot" in a single flow? Many. In real-life, hearing your enemy run parallel to you on the other side of the wall will mean you plan exactly that: sprint to a break in wall, aim, half-squeeze. Yup, the crux is in the ability to half-squeeze. Those who perfect trigger control won't actually fire into thin air, but will still shoot the enemy within a split second (completing the squeeze) should he appear.

But you still want a more realistic game mechanic? More realistic than Frozen Synapse currently offers? Next section. :)

More Realism, More Intuitive Playability

Once again, let's take the act Fire (firing a gun) to be 0.5 second. Let's have 2 opposing soldiers sprinting out of own corners (2 seconds), aim (0.2 second) and fire (0.5 seconds) at each other.

This time, let's do a realistic mechanism that Frozen Synapse doesn't do: Make "aiming" actually mean "faster to fire off a shot". (Gamers had to figure out themselves painstakingly that the act of "aiming" doesn't improve engagement outcome, only the act of "walking slower because of aiming" does.)

X-Com does it. If you're stationary but not aiming, you'll squeeze off the shot slower, less accurate even. If you're stationary and aiming, you'll squeeze off faster and more accurately.

How to do it in our game mechanics? Make the Command Issuance Cost for firing 0.2 second when aiming. Here's the parallel illustration (parallel to previous section).

Aggressive Player takes a go-go-go approach, issues a single command session for first 2 actions, incurring a total cost of 0.5 (command issuance cost) + 2.0 + 0.2 = 2.7 seconds. Turtle Player issues a single sprint (0.5 + 2 = 2.5 seconds).

By the 2.5 second mark, both soldiers have cleared their corners. Aggressive Player cannot issue any command because his soldier's "chain of commands" is 2.7 seconds. Turtle Player's "chain of commands" is only 2.5 seconds, so he can react to the situation (issue a new session of commands).

How does this benefit Aggressive Player? Turtle Player attempts to aim and fire. These commands cost 0.5 + 0.2 + 0.5 = 1.2 seconds, which fires off the shot at the 3.7-second mark. However, he is shot by Aggressive Player at the 3.6-second mark (0.2 catch-up + 0.2 command issuance cost + 0.5 to fire)! Score 1 to Aggressive Player!

How does this benefit Turtle Player? He knows he was slow (due to go-see-go approach). Not possible, since it takes 1 second to sprint back into cover (see previous section). What do we do? Another realistic thing! Introducing the "dive"! Same speed as the sprint (0.5 seconds to return behind cover), but takes much less Command Issuance Cost: just 0.2 seconds. Only drawback is that a dive runs for a full 3 seconds (travel 1 second, recover 2 seconds). In this case, we don't mind, since we'll be doing the "recovering" behind cover. So, Turtle Player initiates a dive that takes 0.2 (command issuance) + 0.5 = 0.7 seconds to reach cover. That makes the dive 0.4 seconds ahead of time! At 3.2-second mark, one player dives back into cover, while at 3.6-second mark, the other player squeezes off a shot. Good save!

Game mechanics sound complicated, but realize that computers are the new "game masters". No humans are required to keep score painstakingly today.

Plus the game becomes intuitive (mirrors real world). Not so in Frozen Synapse. Currently in Frozen Synapse, you would think that remaining stationary plus aiming gives you a faster squeeze-off time than just remaining stationary alone. But alas, the gamers were left out cold to figure that un-intuitive facet of gameplay for themselves. Mode 7 should rename their current "Aim" command to "Face" instead.

Stalemates are easily broken

Let's consider turtle vs turtle. If all both players do is Aim, then it is true that each play session is only 0.7 seconds. So will that lead to slow gameplay? No.

2 scenarios.

If both players are online, gameplay is blazing fast. "I aim (0.7 seconds), you aim (0.7 seconds), I aim (0.7 seconds), you aim (0.7 seconds), I get bored and start sprinting for your behind (2 seconds)".

But what about async play? Next section.

Fine-grained (Turtle) or Coarse-grained (Aggro) control? Your choice, your risks

Frozen Synapse currently gives you only coarse-grained control. Put your human reactions on backseat, press Go-go-go, and hope you guessed your opponent's go-go-go paths corrently.

But what about async play? Will each turn become so fine-grained that each turn only sees soldiers sprinting 14 meters each turn? (A 2-second sprint averages to 14 meters.)

No, typical turns will see players commit as large a time-block as 6-20 seconds. In the least, a typical turn will see a player commit 3 sprint commands (6-second turn), since each sprint is a straight line that covers ground in only 2 seconds. Recall how being aggressive saves you a lot of time, shaves off Command Issuance Cost.

Then you'd think "Oh gosh, 3 sprint commands is more tedious than 1 single move command". But you'll be wrong because of this: planning paralysis. Moreover, the game can automatically break-up your "single move command" into discrete sprint commands.

If you break up the entire move into discrete and easily calculable packets of 2-second sprints, you'll find that you lay out your plan much faster. And if you're into computer science, this is also why serial data processing (1 at a time) is faster than parallel data processing (consider that SATA has overtaken PATA, and USB has overtaken parallel DB-25 ports).

Why commit up to 3 sprint commands? Don't you wanna change your mind after 2 seconds of sprinting? Sure you can. But that'll mean tagging on another 0.5 seconds for Command Issuance Cost. See how turtle play and aggressive play are both viable potent options?

About serial data processing, it's a simple concept, why it works. Focus on the "now", think things through 1 step at a time. Makes for faster and clearer decision-making.

Anyway, I've simplified the game mechanics for illustration. You can actually sprint a minimum of 2 seconds, but a maximum of whatever-knocks-you-out-marathon-length. Also, it takes maybe 1 second to come out of a mid-sprint to respond to contacts, less for a mid-walk. If you're able to extrapolate from the above 2 sentences, you'll begin to understand how this game mechanic mirrors real life and is still as playable (or more) as Frozen Synapse.

In short, have the game pause at vital points for you to choose to spend a 0.5-second Command Issuance Cost to issue new commands. Or for you to choose to continue on. Frozen Synapse currently forces a pause at 5-second marks.

So, both types of play, Turtle and Aggressive, receives due rewards. Turtle play rewards ability to response to any and every enemy action. Aggressive play (chained commands) rewards huge time-savings.

No Pause on Engagement means User Interface gets complex AND buggy!

Yes, the zone control. Focus/Ignore zone interface is buggy. Why? Because Mode 7 had to enhance the user interface to work around problems in game mechanics.

What problems, you ask? Here's the reason why Mode 7 attempted to introduce zones, mostly for Dark (fog-of-war) games.

A machine gunner moves forward. To the front-and-left, is a series of windows (covers) where enemies are possibly headed for. On the front-and-right is a wide open space where the machine gunner's range can advantageously cover. However, the machine gunner moves forward and instantly discovers an enemy at a window on the front-and-left, and attempts to engage him in a sure-lose exchange of bullets. Meanwhile, an enemy that popped out in the front-and-right shoots him down. An ideal outcome would have been for him to kill the enemy on the right, despite being gunned down by that on the left a split second later.

In short, all Dark games in Frozen Synapse were broken because of the inability to "Ignore Zone".

Now you see where we're going? Rather than improving the game mechanics (tougher on the developers), they are "improving" the user-interface (tougher on the gamers).

Clunky Game Mechanics Create Superfluous New Units

So it's impossible to do real feinting in firefights within Frozen Synapse. Intuitively, our real world knowledge would have let us do feints and double-feints, effectively breaking stalemates and camper-lockdowns. Alas, Frozen Synapse doesn't allow any of that. (My feints actually backfire half the time! Ok, here's the vital tip. A feinter will beat non-feinter. A double-feinter will beat a feinter. A non-feinter will beat a double-feinter. Ask for replays to illustrate this! Or just use your maths, it's trivial.)

So what does Mode 7 do? Introduce the Riot Shield unit, which does help break stalemates. Barring a rocket hit, the Riot Shield unit allows a machine gunner to become Superman out in the open!

Chances are that 2 things will happen to that unit: Either it will be scrapped because it can be abused, or it will be ignored completely because it is actually ineffective (waste 1 slot that would have been another gun).

And then Mode 7 would pursue the "arms race" that plagued many "expansions". Create a unit that imbalances the game, nerf it or drop it, or create yet another counter. And the vicious cycle goes on. And the efforts to re-balance the game (unproductive backtracking) occupies the developers more than the efforts to actually IMPROVE the game.

Give me a Riot Shield unit, and I'll have THE perfect feinter: one that doesn't even have to dodge bullets!

So what of Frozen Synapse Now?

As more and more people flock to Frozen Synapse to exercise their tactical muscles, they'll realize there are huge holes in the game mechanics. It's simply not the test of tactics people hoped for.

And then less and less people will be playing Frozen Synapse (read online how many own the game but have left it alone for now).

People will still play because it offers a great game of psychology (second-guessing) and some tactics. Specifically, it's currently 40% tactics, 50% psychology, 10% real urban warfare know-how (leap-frog, cover angles, etc). And that's precisely why I'll play it too! Also, it's the best real-time tactics out there, until I come up with better. :)

But Frozen Synapse will never have many players. Typically just 30-50 players on the best nights. Consider that all do async play, let's multiply that figure by 10... 300-500 players a night, aggressively speaking. Compare that to chess games online, and you'll be wondering why chess has 10 times more audience than Frozen Synapse. Isn't Frozen Synapse supposed to be more fun than chess? Well, you can put down "FIDE rating (chess) of 2300" in your resume, but not Frozen Synapse. Makes sense to invest time improving your chess rather than Frozen Synapse, no? :)

I tried to introduce Frozen Synapse to my wife. But now, I'm thinking I should save her time, and teach her chess instead.

Hope all that explains it well enough for you.

The entire design is actually better explained in a board game, blow by blow. In fact, that's exactly what we do for urban tactics analysis. ;) Only reason why it's never been a popular game is because the board game master needs to handle lots of accounting and calculations. That's where the computer comes in as the game master, right? I did say that the game mechanics need to become more complex, not the user-interface. :)

As I said, nobody wanted a hybrid of real-time tactics and turn-based tactics until Frozen Synapse came up with it. (Actually, some other space ship and missile game in 1995 came up with it first, then Steambirds, then Frozen Synapse. But kudos to the ones who execute it best: Mode 7).

Re: Wait timer does not count down with aim on?

Posted: Sun Apr 06, 2014 1:37 pm
by akimbomidget
I like the concept you mention.

If you are going to be developing this game. I would recommend that you consider making the game engine (not the UI engine) based off a virtual machine like say lua/python/etc... with exposed and editable ruleset for multiplayer.

You may think it's cheatable, but as long as everyone is given a copy of the rules and does their own verification, no one guy can cheat in this game. (everyone send each other their encrypted moves during planning, then sends decryption keys during execution/simulation.)

What this would allow for; is community experimentation in improving the realism of the game (or alternatively the fun aspect). E.g. tweaking the reaction values, or adding mental states (shellshock) etc...