29th May 2020 I just want to caution everyone this is a long guide based on the current mechanics since the Steam Pipe update of 2013. That update had dramatically changed a lot from initial release of the game back in 2009. Valve has recently been pushing pretty big updates[steamdb.info]. This could change a lot again, thus making this guide redundant in the future. And the last thing I want is to waste anybody's time with obsolete information. I'll keep this preface updated should that occur.

18th Sep 2020 So it's The Last Stand update. I had a feeling the navmesh would be changed again. To be completely honest, when I wrote this guide in 2017, it completely slipped my mind that they could potentially update this game again like with the 2013 Steam Pipe integration. So I don't know if I'll update this guide, probably not... obviously I'll take a look at the changes, but I'll leave everything in this guide for historical purposes for now.

6th Jul 2021 Within 10 months since my Sep post, the game had some minor changes and then another major update including navmesh changes recently. In that time, I've been playing with a few players helping them get 20 mins on all maps. So now I've been pondering how I should approach updating this guide. In all honesty I don't think it's the same anymore. There's very minor flexibility. And no room for error... a la... Left 4 Dead 1 survival style. I'll start by blanking out any text in this guide that might be redundant. And then chip away at the other parts.

I've been playing a lot again with the amazing Azimuth. He's been working hard to update the Gold Miners L4D2 Survival Guide which I believe to be more relevant now.

10th Jul 2021 It's also probably worth checking out this website. Just be warned, it's not affiliated with Valve and more than likely owned by an individual collaborating with Gold Miners: http://survival.site.nfoservers.com/DemoArchive/. I can't always check the validity of all links. Also remember, if you download and view demo's your time can be overwritten on your Steam profile statistics for the game. So always watch it in Steam offline mode.

Aside from the game being updated twice again since last night, which theoretically should make updating this guide more difficult. I went proofreading over what I had written very quickly. And surprisingly most of what I have "written" is still highly relevant. So the point of this guide (which I had forgotten) is to try and make you a better player without necessarily following strategies. So get cracking...

If you've made it this far pat yourself on the back...


Originally I wrote this guide out of pure frustration. I was upset that I was playing with a new generation of players that have the luxury of following strategies that had already been defined. The problem was there were many failures and brute forcing due to not understanding more than just the strategies available. As a result, too much of my time has been wasted. Continuing without having fun would be an insult to my former self who'd already wasted a lot of time.

I asked a buddy for help to write this guide. He declined, saying that the present community was toxic and that they didn't deserve it. I got him to at least glance over whatever I had written. He saw my frustration and thanks to him I had hopefully removed most of my negativity. Calm blue ocean.

After I had finished the L4D2 SAVL Series, I feel like this community toxicity had somehow indirectly effected me through other unsuspecting players that I play with that had been indirectly affected themselves. It was enough to make me not want to post this guide. I felt sick to my stomach.

But I've made the decision to release this guide anyways. I've no proof that the community really is toxic other than a gut feeling. If it's toxic and they choose to crucify me based on the contents of this guide, then so be it. I suppose it's worth the price knowing in order to keep my lips sealed next time, should I decide to create something like this again.


This guide has a couple of purposes. First to teach newbies the basics. Second to teach some serious players who have high times because of following strategies to a T or were carried, but have no basic foundation of how to really play. Third to make it clear that record times aren't really records due to uneven playing fields. Fourth a reason to retire by giving a glimpse into my mind and moral code, which will hopefully better the survival community by avoiding dodgy or ambiguous play.


Survival Mode is a team game, don't solo off doing your own thing unless you're an experienced player. Otherwise you'll drag your team down with you. In cases like these, the AI bots are better than those human players. This is why getting a mere 10 mins can be difficult in a public game.

Surprisingly some kids who enter a survival game, play a lot better than their adult counterpart. And often some adults act childish, go figure...


When you're green, don't heal up or take pills. If I had a nickel for every time that happened...


Don't suck up to or glorify players just because they have high times. They could be glitched or the player could have been carried by veteran players. The only way to know how good someone is, is by spectating them and seeing their style. They'll be many different styles for each individual that showcases their strengths (and weaknesses).


If it helps and I'll say it right to the camera, STOP IT!


You should always use your primary weapon, especially if there's a Tank or Special Infected in your view. There's nothing to be gained by using pistols only. In fact if you do, you increase your chances of a leprechaun knocking on your bedroom window at night.


By default, press C button to allow your voice to be heard. If you're a 10 year old boy, some people might mistake you for a girl. And if you're a woman, you might be mistaken for a 10 year old boy. This phenomenon could be due to poor microphone quality, or maybe the player at the other end just doesn't have a good ear for hearing the difference. But either way it's nobody's fault.


If you're dead, you'll be spectating other players that are still alive. And if it seems like a walk in the park for them, there's a good chance you're spectating experienced players. In this case don't constantly ask them to defib you, because they might not want to then. Give them 5 mins.

At any one time on a map there should be 50 Common Infected (if randomly Jimmy Gibbs commons appears, call a restart), 7 Special Infected (any combination) and 2 Tanks (very rare case could be 3, or extremely rare 4 or more, it has happened).


Over the years since 2009, Common Infected behaviour has changed. They were more consistent prior to the 2013 SteamPipe update. But there are some typical behaviours you should be wary of. When a horde approches you, a single one (or two) will try to swing around to your backside. This gives you the sensation of being stuck. The other is being straight up boxed in.


There's nothing to it. All Special Infected will make distinctive sounds indicating a direction of threat. If you have a surround sound setup, this is an advantage. But headphones are just as good. If you rotate your character often, you can listen in which direction the threat is coming from.


This requires experience or pure skill. It's smartly reacting to a given scenario. For example, lets say the best player in the team ends up getting incapped and is being beaten by Common Infected in a difficult spot to reach, far away from the main holding spot. What do you do? Now think about that for a sec... if your only initial thought is, "Let's save him/her immediately!", think again.


Don't panic with looking at the timer. Even experienced players make this mistake. They'll think, "Oh we're going to get 30." then they panic and die at 29. Or, "Woo Hoo, I'm going to beat my personal best!" then bam, dead a second too early. It's because they begin to focus more on the timer, mentally counting the seconds rather than focusing on the action. Time should only be used as hints of knowing when there are big breaks, e.g: 4 mins, 6 mins, 9 mins, 12 mins, 15 mins, 30 mins... etc.


For survival, there are only 6 primary weapons you should choose from depending on which map you select. Choose the weapon you feel like you'll be most comfortable at using. But for those who are inexperienced with survival mode, this is the order in which you should be taking:

1. AK-47
   
2. Military Sniper Rifle
   
3. M-16 or Desert Rifle
   
4. Auto-Shotgun or Combat Shotgun

Whilst explaining why is beyond the scope of this guide, this doesn't mean shotguns are not as effective as the AK-47. Primary weapon effectiveness depends on a number of factors including your knowledge of all weapons (damage per shot, reload speed), weapons other players are using, your accuracy, whether the map has a wide open landscape (more common) or closely spaced, whether you'll be running and gunning (or not) and whether a particular strategy is being implemented.


An experienced player will usually take a sharp melee. If you fear it will be difficult for the other team members to pick you up whilst incapped, take a Magnum. A Magnum is also good as extra firepower when your primary weapon magazine is empty but you still need to shoot something. The Magnum also pairs up well with a shotgun. I wouldn't recommend pistols, unless you're implementing a strategy that specifically requires it.


Does it make you scratch your head when you see players throwing 4 pipe bombs, 3 molotovs, 2 boomer biles and blowing up 8 gasoline cans in less than 2 mins? True story. A lot of the times, it's just better to shoot. By the time you switch over to a throwable and you're just thinking about where or when to throw it, you could have emptied a full magazine of your primary weapon. Having said that, here are some general guidelines:

Molotov You can use one on a Tank or for back burning an area so you can focus shooting in a more important direction.

Gasoline For inexperienced teams, I wouldn't recommend collecting up a pile of them somewhere. They'll just get shot up by some n00b in the team. Instead spread them out or just leave them where they are. Because the chances are some n00b will still go around shooting them all up within 2 mins. They are used for burning Tanks mainly and at the same time back burning an area, much like the Molotov. A fun trivia; if you pick up a gasoline can, drop it and then melee it before the round starts, they will blow up. But after you start the round, this will no longer happen.

Propane These are mostly useless for an inexperienced player. But they are used by experienced players to stumble oncoming Tanks.

Pipe Bombs When a player gets boxed in by Common Infected for more than a second and there's a Tank or other SI nearby that can cause them serious damage, use one.

Boomer Bile These are much like a better version of a Pipe Bomb. But they are mainly used when someone has died in a really difficult spot to defib. They are also used for complicated supply runs that are far away from a holding spot. For maps like Traincar and Port Sacrifice they are randomly dropped by Common Infected and are used at certain intervals to prolongue the survival round.

I didn't really think about this section. It's more like a mash up of things that may be important but I'm not sure whether I have mentioned about it yet.


todo


If you happen to walk into a narrow door, or maybe up the stairs as you're lapping in Mall Atrium, you can blindly melee to stop a potential Jockey or Hunter landing on your head. Think of this like a precaution.


As you're moving (whether playing Smash TV style or implementing a lapping strategy) you're ahead of time. Having red latency may give you a slight advantage from time to time whilst you're on the move as the server will see you irregularly teleporting from position to position. However once the Common Infected catch up and box you in, you become stationary and it becomes really difficult to melee your way out because the server is now ahead of you. This is prevalent when playing on a server overseas. Of course with green latency this is almost transparent as you'll almost always be able to melee yourself out of a boxed-in situation.


Because the players are the ones with the guns, it can be advantagous to hide behind a slim pole to evade a smoker tongue. As you get closer to the pole, it becomes wider on your screen.

By being up against some wall, you also have more control over Common Infected beating you due to less surface area surrounding the player.


A griefer is someone who enters a game (usually public) and either shoots another player, uses up consumables or just commits suicide. This is intentionally sabotaging a round. The problem with this is, almost all public games aren't even worth a damn sabotaging and every single serious or semi-serious game out there is friends-only with a full team already. Furthermore, there's no point for a griefer to suicide on maps that incorporate a lapping strategy like Mall Atrium. If there are experienced players there, they might actually be glad as they can lap for fun which could potentially go on from 30 mins to over an hour.


Setting a Tank on fire will not make them move faster. Setting a Hunter on fire will not cause them to do rapid damage when they pounce on a player. This confusion stems from L4D1 attributes.


When a Charger charges into a wall with a player nearby, that player will get stumbled. To avoid this, the player should jump up in the air just before the moment of impact.


It's pretty rare, but there are god frames in the game. Usually the animation just after someone has been picked up from being incapped but before they can defend themselves. If they're attacked during this time, usually they won't loose health.


Some weapons like the useless Silenced SMG slows the Tank's movement speed. It's still useless.

No, you're not. If you've been accepted to play some serious rounds among veteran players, you better damn well know these basics! Because they sure will know whether you don't!


Like Counter-Strike, with certain weapons and wall structures you can shoot through them, thus killing. This is usually any non-imported 3D model, i.e BSP (Binary Space Partitioning) brushes that was used to create simple geometry that lay the basic foundation of a map.


Jumping out of spitter goo will still cause you damage. You're better off just walking out of it quickly.


Move your mouse rapidly. You're doing it right if you're not really being ridden anywhere. By doing this, you'll stay put and your team mates will have better chances to save you.


Ideally you shouldn't be damaging other team mates. However it's okay doing a bit of damage like killing Common Infected surrounding them if it saves them from getting smacked by a nearby Tank.


Wanting 100 health by 10 mins doesn't necessarily mean you're a good player. In fact you might be making things more difficult for the other players by being greedy. If however you happen to still have 100HP but you're playing hard, then that's okay. For example, in the screenshot below phoenix_advance was holding the front of the highway where everything comes in. It wasn't until 21 mins when he realised, "Oh Gee, I still have 100HP and I'm not hiding behind anyone."


 

It seems daunting to learn but is well worth it. It's a bit of a risk if you botch it, but very convinient if you don't. With green ping you need a sharp melee weapon like sword, axe or crowbar. Listen out for when the smoker is about to release its tongue. When you see it come out, take a swipe. You don't even have to aim for the tongue. You might have to wait a few micro seconds longer if you're being smoked from a long distance.


If you get some shots into a Charger but it doesn't die and starts charging instead, strafe out of its way and swipe at it with your melee. Don't worry if it doesn't die. What's important in survival mode is playing it safe and doing your part. It'll make it easier for other team mates to finish the Charger off. Sometimes a second player may even complete the leveling for you by taking a second swipe at the Charger if it's still charging.


Given a hectic scenario, if you wound a Hunter it will jump away. This gives you time to kill something else of higher priority and then focusing on the Hunter when it comes back. Alternatively you can skeet a hunter (mid-air kill), thus conserving ammunition and time. The video below will show some examples of skeeting a Hunter. How to do that exactly is beyond the scope of this guide.



Shoot it down! It's free target practice and you may even save another player's health.


Boomers are like zero damage propane, they can stumble a nearby Tank. Often players (even skilled players) kill Boomer's prematurely.

The goal of this section is to explain commonly used console variables in layman terms. Reading best configurations in forums and listening to technically minded people will make your head itch. It's worth mentioning that by explaining in layman terms, a lot of technical details may get lost.

If you enable developer console in the options and press ~ you can enter these commands. Alternatively you can create autoexec.cfg in steamapps\common\Left 4 Dead 2\left4dead2\cfg and place the commands in there. autoexec.cfg is just a text document with a .cfg file extension.


Just like ADSL2+ has a max theoretical downspeed of 24 Megabits per second, the term "tickrate" for a server is similar. At the time of writing this, L4D2 Valve official servers have a tickrate of 30. So realistically it may be more like 29.8 and fluctuate for the player depending on network conditions.

Each server update sent to the client is a single tick, which is akin to key frames in the film world. A tick is a snapshot of how the server sees (typically) positions or angles of entities shared between the client and server in the game world. Ticks are evenly spread out to make game time consistent across different computer hardware. Thus the length of a single tick can be calculated.

length of a single tick in miliseconds = 1 second / server tickrate

For our L4D2 example, this means 3 ticks every 100 miliseconds. We also assume the player isn't moronic enough to cap their update rate to something below the server tickrate. For example using cl_updaterate 20 would mean the moron would get 1 tick less every 100 miliseconds.

There are two ways an interpolation value can be set. And only the highest value out of these two ways is selected. One way is by using the cl_interp command. At the time of writing this for L4D2, the default value is 0.1 which gives you a lerp of 100 miliseconds. The other way is by taking the value of cl_interp_ratio (default 2) and divide by the value of cl_updaterate (default 64). In the unlikely case that the interpolation value is abnormally high, it'll be capped to 250 miliseconds, which is about the average human click reaction speed to visual stimulus. You'll notice the lag since the human brain can process images that the eyes can see for as little as 13 miliseconds[news.mit.edu].


Choosing an interpolation value depends on whether you want to see smooth animation with a slight delay (higher value) or choppy animation with negligable to no delay (lower value). This delay is artifical and only happens on the client-side. So it's worth mentioning that the lower the interpolation value, the closer the hitbox will be to the character model. This poses no advantage or disadvantage. During online play the hitbox will always be ahead in time of a moving character model and what really matters is mastering the distance between the two.


This is more of an opinion. There are various reasons for choppy animation. The most likely being when an entity is moving faster and unpredictably in proportion to what the tickrate can handle for smooth animation. A distant example of this would be playing a slow paced game like Hitman: Absolution on outdated hardware that gives you an average of 15 frames per second. Alarmed? Well don't be, because it's smooth enough to play comfortably. But if you were to try a modern fast paced first-person shooter on outdated hardware, 15 frames per second would be choppy.

Bad network conditions (typically network congestion) can lower the amount of updates a player receives. If for example a player had a packet loss of 25 and choke of 50, they would be seeing teleporting of entities rather than choppy animation. Using the default cl_interp 0.1 would not save a player from this teleporting effect, as it was designed for a loss of 1. See Impact and management.


Playing with high latency, like on a server in some other part of the world, does not necessarily mean bad network conditions. Animation can still work smoothly. But it does imply that since more hops are taken, there's more chances for something to go wrong along the packet's journey. Latency can be found by pressing TAB in-game. It'll either be in the green, orange or red zone.

This is more of a side note. From The Orange Box onwards, it seems almost all packets are well under the Ethernet v2 MTU size of 1500 bytes. Given the sizes that can be observed by using net_graph 4 it's unlikely that compression is being used. Packets can arrive in different order, delayed or the payload may be split up with another packet. This is why the definition of a tick is decoupled from the definition of a network packet, they're not the same thing. Although a packet may often contain a single data payload for an entire tick.


The rate value is in bytes per second. There are 131,072 Bytes in a Megabit. Where Megabits (Mbps) is your download speed which can be found using something like speedtest.net by Ookla, we can perform the following calculation:

rate = Megabits x 131072

At the time of writing this, there is a max rate of 786432 which is a 6Mbps connection. Of course if your download speed is anything below, it wouldn't be a good idea to set your rate to the max of your download speed.


There shouldn't be any issues using a high rate, although it may be pointless if the game doesn't even allow for this. The best way to check is using net_graph 4 and checking the k/s values for the In and Out. These represent network packets in kilobits per second. For L4D2 survival mode it's pretty rare to see anything close to 20 k/s for In and 10 k/s for Out. It's typically something like 14 k/s for In and 4 k/s for Out. After doing simple conversions from kilobytes to bytes you'll find that people who have recommened something like rate 25000 are correct as it's probably more than enough for L4D2 survival mode. Using a rate like 30000, 50000, 60000, 128000 as recomended by many old forum posts and YouTube videos won't really give much benefit as they're all above the minimum of what's required and well under a 1.5 Mbps connection.

To put this in historical context, 1.5 Mbps was common in Australia from 2000 to 2006. Despite being a first world country, Australia is shamefully slow to deploy upgrades to Internet infrastructure.


So cl_cmdrate represents the amount of updates you send the server. It has the same default value of 64 like cl_updaterate which represents the amount of updates you want the server to send you per second. Because these values are similar there is often confusion as to what values are good to use. Whilst Valve themselves have suggested no more than a value of 66, see Optimisations, the truth is it really depends on server configuration limits and client connection specifications.

 

If a team were to play on a server that unlocked the tickrate to 64 or higher, it would be unfair for others who play on servers with the default tickrate of 30. Calling it cheating is debateable, but it's definately a completely different league and records gained this way should be stated as such.

This section has a similar purpose to Part A, but will discuss more straight forward tweaks. A lot of commands in this section were learnt from Kaizen's performance config v1.1 (Nov 21, 2009). I'll do my best to filter out commands to only the most practical and useful, so you can build your own.

All these commands are shoved into an autoexec.cfg file. This is just a new text document that you create. You save the file name as autoexec and then change the file extension to cfg. Then you place the file in steamapps\common\Left 4 Dead 2\left4dead2\cfg along with other cfg files.


By setting the following command to 1, if you press ~ the console will open up. 0 will disable. You can also find this in the options menu in-game.

con_enable 1

Setting the following command to 1 pre-loads levels entirely into RAM. So if you have plenty of RAM to spare, this should always be set to 1 regardless of whether there really is any observable benefit or not. A lot of people will attribute this command to gaining more consistent FPS, which is true from a basic view. Inconsistencies mainly arise when you try playing a game that is way ahead of what your PC specs can handle. L4D2 requirements are that of the latest PC you could build yourself back in 2007 (2 years before the game came out). So you're not likely going to spot a difference especially in 2017, but from a technical perspective, you'd want this turned on.

cl_forcepreload 1

If you don't want custom WAV or MP3 downloaded from the server set to "nosounds". Otherwise set to "all" if you want everything downloaded and "none" if you don't want any sound downloads.

cl_downloadfilter nosounds

Set to 0 to stop any custom downloads from servers and 1 to enable. If you have it disabled and you forget about this whilst trying to join something custom, it may cause you some frustration.

cl_allowdownload 0

lol, not sure if this really works or not anymore. But it's nice to have tips disabled.

cl_autohelp 0

Disable crosshair with 0 or enable with 1. This is usually good if you're making an intro video (for YouTube) and you need to remove the crosshair.

crosshair 1

Transparency 0 to 255, where default is 255. There's probably no point in changing this. As it may be distracting. The point of a crosshair is to see it overlap what you're aiming at.

cl_crosshair_alpha 255

The following allows you to customize your crosshair color with RGB values from 0 to 255. The default neutral colour suits L4D2 in general for all maps. There's probably no point in changing these values either. Let's say you decide to have a sharp green color that's clearly visible. Your focus might be too much on the crosshair then, rather than what the crosshair is overlapping on.

cl_crosshair_red 138
cl_crosshair_green 182
cl_crosshair_blue 220

Changing the following crosshair attributes yeilds no benefits whatsoever. What's important is understanding the bullet spread of your weapons. The default crosshair gives you a hint of this. But once you learn the bullet spread of all weapons, you may want something more minimal. Changing thickness is where you might find a personal tweak. Default is 2.

cl_crosshair_thickness 1.2

Disable dynamic crosshair. This is where you may slightly boost FPS if you have a very poor gaming rig. But as I've mentioned before, it's 2017 now. Most PC's should be able to handle L4D2 quite easily now with a consistent 60 FPS. Back inbetween 2009 and 2013, not so for everyone. Default 1.

cl_crosshair_dynamic 0

Enable crosshair in spectator mode. This serves as a point of reference to the middle of the screen. Spectators see the world without lag compensation.

cl_observercrosshair 1

Change survivor model FOV. Default is 52. For 16:9 ratios 68 is pretty good. For 16:10 try 72. And for Quake style use 180. There's no advantage to using Quake style, for example you may find it difficult to know whether you have thrown a pipebomb, molotov or bile as you cannot see the animation of when a throwable has been thrown. As to seeing more on the bottom right of the screen, is more than likely a placebo effect. A Jockey or creeping Hunter will still get you.

cl_viewmodelfovsurvivor 52

So there's this story in L4D1 survival where a player used some kind of mat hack to insanely make all wall textures extremely bright like as if there was no darkness. The only thing close to describe this is similar to when developing a map for L4D and everything in the Hammer editor is bright. Well somehow this dude made things even brighter than that. This is obviously cheating. The walls in his screenshot were almost pale white. But these are different commands with reasonable values for L4D2. In fact the values match closely with what is recommended for LCD monitors. The defaults are 1 for mat_bloom_scalefactor_scalar and 0 for mat_monitorgamma_tv_enabled.

mat_bloom_scalefactor_scalar 0
mat_monitorgamma_tv_enabled 1.8

Once again for someone with a poor gaming rig, it may help to disable eyeball movement for character models. These tweaks though are really minimal. Don't expect to get 10 more FPS with all tweaks combined. You're better off upgrading your graphics card. Defaults are 1.

r_eyemove 0
r_eyesize 0
r_eyeshift_z 0
r_eyeshift_y 0
r_eyeshift_x 0

If you wear headphones or have stereo speakers, this supposedly enhances with virtual surround sound. Default is 0. Obviously if you have real surround, then you don't need this.

dsp_enhance_stereo 1

Reduces delay in sound. It doesn't really yield any benefit. What kind of matters is that you know the timing of the sound. Default is 0.1, if you go too low it'll cause crackling which is undesired.

snd_mixahead 0.05

Removes debris cluttering the screen. Default is 15.

func_break_max_pieces 0

For more information check out these links. Don't take everything to heart though. Large lists like these aren't always maintained by testing as to what can be considered cheating and what isn't.

Console Command List
List of L4D2 Cvars
List of CS:GO Cvars


There's a whole bunch of launch options, but it's probably a good idea only to use the most minimal, in this case the -novid option which skips the Valve intro video when launching the game. In Kaizen's config, there was some really good explanation of launch parameter options like -refresh, -heapsize, -high, -noforcemaccel, -noforcemparms, -noipx and -nojoy. But these are best left untouched as you're not likely to see any tangible benefit. And it may even do more harm than good. Everything else should be configued within the menu options in-game.

This section is intentionally separate from Game Tweaks - Part A. There may be confusion between the two due to same terminologies being used (but in different context) and empirical coincidences.

There is an sm_cvar command on the server side called nb_update_frequency which may have been tuned to compliment the default for cl_interp on the client side. It's classed as a cheat, although to some that may be debateable, regardless it'll help explain some things.

So the nb in nb_update_frequency stands for NextBot which is the AI system. The command value if set to 0.1 will mean AI tasks will be performed every 100 miliseconds. This artificial delay is to reduce server CPU usage. If you recall the distant Hitman: Absolution example from Game Tweaks - Part A, it also means that Valve has quite possibly tuned nb_update_frequency to make NextBot as smooth as tolerable. This is why if you use a low lerp, you'll see Common Infected moving choppy. It's because you're seeing the most recent game update from the server, but the server itself has imposed an artificial delay. Also the fact the Common Infected move pretty fast and sporadically. If you set nb_update_frequency 0 AI tasks will be without delay at the cost of higher CPU utilization.

So something really important to mention here as it was ambigouous in Game Tweaks - Part A is the definition of "artificial delay". This is definately not the same as network latency delay. Artificial delay is more consistent as it follows D/D/1 in queuing theory[www.cmg.org]. Whereas the Internet in general follows more along the lines of M/M/1. Also latency is an average, which is not a good precise representation in terms of metrics. Furthermore depending on scenario, since overall latency has been calculated in a bidirectional manner (we can infer this from the ping) the numerical value of latency will really be half that much (approximately) from client to server.

When a client receives an update from the server, it doesn't necessarily mean each is always taking half of latency to arrive. For example, like whether the server had to choke back a few updates. Think of this like a tap fitted with a 2-way adaptor. In one way we'll attach a hose of 50 meters length and the other 120 meters. When you turn the tap on, water (the packets) will flow and come out of the shorter hose first. 70 meters later the longer hose will have water coming out. The point being made here is the flow of water is continuous. It's not like as if we have to wait 120 meters on the longer hose before the next fresh water that we haven't seen comes out... it was already travelling along the hose. This is also another very simplistic D/D/1 example that doesn't best represent the Internet in general. But the flow is what's important to know.

Let's pretend all players connected to this game server had 0 miliseconds of interpolation. Each player is seeing an event that occured in the past. The Source engine allows everyone to see their own different delayed view concurrently depending on network latency. This ensures continuous gameplay. Other engines like the Buddha game engine for Iron Brigade seem to synchronize everyone together, the result being the game crawling down to a halt for everyone at the same time until everyone is on the same page seeing the same thing.

This is where lag compensation comes into play. It's essentially a 1 second (1000 miliseconds) buffer of past events on the server side. A player sends a user command to the server and this is checked with what's in the buffer to determine whether the hit was accurate or not.

But let's say two players decide to shoot poor Mr. Boomer. And they both land accurate headshots. If one player had 50 latency and the other had 120 latency, the player with the lower latency would win, whilst the user command of the other player would be ignored.

Because in survival mode we're versing the AI, there's no benefit to playing with red latency. In fact it can be detrimental to other players with green latency. Take a verses game for example, you may have seen someone yell out "What a laggy mofo!" because the lagging player on the other team is hard to hit. Similarly in survival, hit registration may be poor with green latency players if they are playing with someone with red latency. This may happen especially during a busy scene and the red latency player doesn't necessarily have to be spraying and praying in the same direction as a green latency player, which implies it may be a calculation or timing issue in regards to CPU utilization.

Some common reasons for bad latency is playing on a server located overseas, heavily downloading or uploading (like to YouTube) on the same network you're playing on (although having a fibre to the premises connection and a $2000 Cisco router may help aleviate this bottleneck) or a player (like a griefer) may do this with intention using software hacks. The second point aforementioned is actually quite common. For example, on the same home network, if someone is playing a game like For Honor you'll get bad hit registration even if net_graph may report that everything's fine.

This is more of a personal opinion. Locally hosting a game has its downsides compared to a purchased server in some data center. The effects feel like bad hit registration and responsiveness. You can be playing from Australia on a Canadian, American, Japanese, New Zealander or even a locally hosted dedicated Australian server, but the effects are consistent and can be differentiated from other things like geographical latency. The major reason is to do with the connection to someones premesis. It could be congested at the exchange or the quality of the line isn't on par with dedicated fibre (like Hybrid Fibre Coaxial). It adds additional latency for the other three players connecting to the host. The host themselves may not have a powerful enough CPU to perform calculations in addition to other running tasks on the operating system. Or the host is on a home network where several people are connected and streaming HD YouTube videos. In the end you'll almost always find that there's packet loss and a higher var (the closer to zero, the better).

Connecting from Australia. Above is a dedicated home server in Australia. The connection is ADSL2+ via a second telephone line. The round was started. Note the packet loss and high var. The var was 1 to 4 before the round started. This is bad.

Above is a computer in New Zealand that has locally started a game via lobby. The round has not been started yet. It's a fibre connection with possibly multiple people on the same home network streaming HD YouTube videos. The var was inbetween 0.60 to 0.90 approx. This is quite good.

There are a ton of bugs and exploits that players begin to discover over time. In this section we're not going to talk about being launched to a Jesus spot, because that is plain cheating. Nor are we going to classify whether something should be classed as an exploit or not.

In the very beginning when getting a mere 20 mins was difficult, juggling items away from their original spawn locations was common. When times began to more than double (roughly sometime before the 2013 SteamPipe update), it was decided that juggling items should be stopped. So stop it.

Weapon Spawn Moving

When The Passing DLC was released, it was discovered that weapon spawns could be moved in its maps. And no thanks to Valve, they still can be. But it was decided sometime in late 2011 to have it stopped as it was exploitative. The interesting thing about this was from the moment it was discovered, the legitimacy of it was always being questioned despite being used.

Tanks have to climb over explosive or fire ammo that had been deployed just a bit off the ground (like onto a moveable barrel) in its AI path. This would give an extra 2-3 seconds delay. However as time grew on, some players began to discover it was possible to block Tanks permanently if ammo packs were deployed with very specific placements. This was obviously exploitative. A good example of this was tested on the map Sugar Mill, where ammo pack blocking ended mid-2011.

Idle Bug

With usage of the bind command in the developer console, one can bind going idle. If you repeatedly tap this button, you can flick idle on and off quickly and repetitively. This has the effect of becoming immune to damage. For example if a Charger is slamming you on the ground repeatedly, the Idle Bug will allow you to keep your health. You should vote kick someone that does this.

This is more of a personal opinion. If you have somehow mastered the skill of bunny-hopping then you probably deserve to bhop. However, I'm aware that there are scripts that make bunny-hopping easy. In this case that would be cheating. But for the greater good of the community and to prevent ambiguity, you should refrain from bunny-hopping when going for a serious time that can be listed as a record. This prevents the playing field from being more uneven.

Coming in with the 2013 SteamPipe update, now it's even easier to install mods. Unfortunately mods are a thing that has to be reviewed on a per mod basis. The "Silent Weapons Mod" for example caused a great stir in the community. As of writing, the outcome was it got generally accepted. This has caused certain players to stop playing as they saw this mod as cheating. For the remaining players, those that don't accept it don't use it. But know that in order to keep playing, they have to mix in with players who use the mod and put up with this outcome. To give a guage of workshop items that cross the line, is having transparent/opaque walls and brightly colour coded (textured) Common Infected and Special Infected.

This is something that carried on over from Left 4 Dead 1 survival mode. Using a certain strategy on The Gas Station map, one could block Common Infected climbing up the scissor lift ladder. It's less prominent in Left 4 Dead 2 due to the less campy nature of gameplay mechanics. But nevertheless the concept works the same and this should not be done.

Spawn Shutdown

Not to be confused with Spawn Blocking, although it may utilize that. After the 2013 SteamPipe update, it was discovered on Mall Atrium, one could completely shutdown a spawn by standing in a particular area for exactly 14 mins 30 seconds. This is extremely exploitative. It may not work the same on each map, but shutting down a spawn point with intention... as Trump would say, STOP IT!

This is the art of getting a Tank to move back and fourth, trying to hit you around some 3D imported static model. To normal public players, this might be fine as you won't master it immediately. But to anyone who is competing for a legitimate record, you should really avoid this at all costs! The most notably debated of all this is camping on the fence in Bus Depot. The outcome was to not use it. The boxes below in the alcove in the warehouse of Port Passing has also been considered a grey-area.

This is more of a personal opinion. As of writing this, the following tactics are accepted but to a degree works conceptually simmilar to the fence on Bus Depot. First, the ledge trick on Rooftop. Second, dancing with the Tank around the window hole on the top floor of Mall Atrium. The saferoom window hole is okay if implementing the saferoom hold strategy as it's core to the strategy. Otherwise they'd be no saferoom hold strategy. Third, tricking the Tank to climb up for the person on the big vent on the roof of Burger Tank. Lastly anything simmilar in any map that allows you to stand on any 3D imported model and bait the Tank, for example: The white van in Port Passing/Sacrifice, the large aircon unit on the roof of Motel, hedges on Stadium Gate... etc.

This somewhat relates to the issue with Tank Dancing. But is rather the manuipulation of all AI pathing for any infected in general with premeditated intent. Some players have played for a long time and really understand how AI pathing works to their advantage. It's hard to unlearn what has been learnt, so this is just a statement for now.

*opinion* I can't remember how this ones goes. Something to do with entering "jointeam 1" and "jointeam 2" into console. But the point is, I saw this being done on an Official Valve server. Not sure if it's something that is still allowed. Obviously cheating.

This section explains issues beyond the action of intent of a player, yet adversely affects what would otherwise be considered a normal round.


When you have consumed pills or a shot, your health should slowly count down to whatever solid bar of health you still have left. Sometimes this doesn't happen. This should be tested at the very beginning of any gaming session. Simply injure yourself and consume some pills. Or shoot another player and get them to take some pills. If health doesn't count down, time to find another server. If you don't test, you risk not noticing or become lazy to stop playing, thus permitting the issue.

Far Spawns

Post 2013 SteamPipe update. It's random and can happen on any map. This is what makes the game really easy. In a game with close spawns, it's very hectic. No matter where you go a Special Infected will be waiting for you around the nearest corner. With far spawns, they can spawn easily twice the distance away. This doesn't impede kill count as you can still shoot and kill Special Infected far away. Which also means Special Infected kills per minute may not be a good measurement of difficulty. Unfortunately players can be playing legitimately and thus these can pass as records. What about the teams who played with close spawns? It's like comparing 24 carat gold with 10 carat gold.


Special Infected get stuck on maps from time to time. It isn't the duty of players to go out of their way to kill stuck Special Infected. And the amount of times will be different for each team. A good example of this is Generator Room. Let's say it takes 20 mins for all Special Infected to get stuck. Now apart from the Tank's and Common Infected, it's relatively safe to go down and throw a Molotov at the stuck. For another team it could be 50 mins before everything gets stuck, with only 3 Special Infected coming in from 20 mins onwards.


Sometimes when joining a server you might see an extra kit or pill left behind on the floor from what seems to be a fallen survivor player. *opinion* I haven't researched into why this happens sometimes, it's like as if the players quit mid game, but as the server resets, the extra items on their characters drop to the floor instead of being deleted.

To the L4D2 survival community,

For those that contributed good ideas and brought issues to attention.

For those that try hard to form a full team where there usually isn't any.

For those loyal friends who stick around and aren't just time grabbers.

You guys know who you are...