Yahn Bernier It seems like you changes a lot of the network stuff from what was originally with the Quake 2 engine. Will the average player see a large improvement from this? Yes. We have a lot more functionality in the engine, with our scripted sequences, skeletal animation system and the like, and so we had to come up with a networking solution that fit the state of our technology. Our goals are similar to those that John Carmack was trying to improve upon with Quakeworld and Quake II, i.e., to account for channel latency and to minimize client <-> server data traffic. We also had to balance those goals with the reality of our technology. What we have implemented is a pretty robust physics model for our players that is used by our client-side prediction models to determine movement outcomes on lagged connections. The bottom line is that a lagged connection is a lagged connection. Our design goal with Half-Life has always been to make the player feel totally involved in the game, even on a lagged connection. One way to do this is always to provide real-time feedback. Client-side prediction fits well into this model, since, in its most aggressive mode (in Half-Life players can tweak aspects of the feel), even on a 1500 ping (totally unplayable - don't try this at home) connection, you can get instant response to your movement controls. Of course, we still rely on an authoritative server, so certain things like weapon firing rely on receipt of server updates. Also, with respect to data traffic, we have taken a close look at compression of the data stream that we send to the client in order to minimize traffic over the stream. This should allow for more responsive connections, as well as allowing more players to coexist on a server. We also have designed certain "lag friendly" weapons such as the Snark so that players on lagged connections can still cause havoc in multiplayer games. Besides Deathmatch, Coop, and possibly Team Fortress 2, are you planning on including any other Multiplayer games? If so, what are they? Nothing has been announced, but we have several things planned. What is your Favorite thing to do in a multiplayer game? Let loose a bunch of Snarks (nasty little bioweapons) on an unsuspecting player and admire their chomping on him/her. With the Spray Paint, can players actually paint what they want in the level, or is everything done outside the game and imported in? For example, can I just stop fighting and draw a picture on the wall? It depends on the images you have loaded in. We will be using our resource propagation technology to transmit squad logos and other personalized logos/decals to other peers in a multiplayer game. Thus, spraying these logos on walls will be visible to all players. It would be trivial to have a solid color, smallish logo that you just use to draw on walls with. I'm pretty curious to see what kinds of outrageous use this will get put to by the more creative/sadistic players out there. And no, you cannot spray paint other players. In the Half-Life chat you guys talked about beam effects. Would it be possible to create something like a laser that can burn through walls? That would be great for DM. It would be possible to create a beam that scorches walls. The "through walls" portion of your question is a bit trickier. I think a pretty convincing version of what you are describing could be generated by a mod-maker. With the decal system putting blood on the walls, it probably gets pretty messy after a while. Will all the decals effect multiplayer performance?. That's one of the tuning issues we have yet to resolve. We have a ton of control over the decaling system and will probably make certain adjustments during the multiplayer game (most likely settings that the server operator or even the player can configure about how many decals they want to see). My recollection is that decaling only causes a short message from server to client when it is first applied and after that point there is no network traffic associated with the decal . It is just rendered on the client's screen. Of course, this means that if you join a fairly lengthy multiplayer game that is in progress, you will not see the "evidence" of the carnage, you will only get the decals that are emitted from that point on. There is not really much server <-> client network bandwidth used for this, but the client will have to render all of the decals locally and that could eventually become a performance drag on the local client's framerate. It would take a lot of decals to get there. "Thanos" said " HL multiplayer goes beyond just throwing spawn points in single player levels." in the #3dgaming chat. Can you elaborate on this a little? See answer to # 2. Does multiplayer Half-Life benefit from global entities in any way? No. Because multiplayer games are restricted to a single level, the use of globally named entities does not benefit those games. Of course, globally named entities are one of the coolest features about the single player game and allow us to provide a huge, seemless world for the player to experience.