Myths and Facts about Wifi/Networking

[IWuvGeno]

This is a post I made on AllisBrawl and thought placing it here would be helpful as well:

I've seen a post either here or on SWF dealing with this issue, but there are still many people who are unclear about the majority of these issues. I'll try to answer these matters based on how often I hear them.


My friend is only 1 to 2 hours away. Why am I lagging?
As a note, this isn't the most scientific measurement of distance, but most people use it. Your ISP (Internet Service Provider) needs to route a connection between you and your friend. This routing requires what is called 'server hops' - or, your connection will take a certain number of stops along the way (think of it like a multi-stop flight). This can make your latency ridiculously high if the server hop count is high, even over the same distance as someone who doesn't get a lot of hops. Generally, when people use two different ISP's, it will make more hops. Certain ISP's are worse than others and may make worse routes, resulting in horrendous lag.

I get no lag when playing people across the country!
Since your latency (response time) degrades greatly over distance, this is simply impossible with current technology. The more distance you and your opponent have between you, the higher your latency (or lag) will be. It might still be playable by your standards, no one is arguing that. Casual games can be played in fairly laggy conditions. From my research, you get about 5ms delay per 100 miles on a decent route. I'll explain how this is important later.


My connection is super fast!
First, I'll dispel the myth of 'super fast' connections. With any connection there are two main factors: Latency and Bandwidth. Bandwidth is how much data you can transfer and is barely important in playing online games. As long as you don't download while playing and get more than 5k uploading and downloading, you're golden. Latency, on the other hand, is determined by server hops and the quality of material used by your network. Some ISP's are worse than others, but most broadband services are around the same quality. AT&T's DSL service is the only bad one in my area, and if you game, avoid this ISP like the plague.


Why don't I lag as badly on other games?
Most internet games use certain tactics to reduce the necessity of super low latency (called prediction), unlike Super Smash Bros Brawl. As you play a game, if you've ever noticed a player suddenly teleporting backwards (rubber banding) or stopping and then speeding way up, you've noticed the results of prediction code. This type of coding wouldn't work for a game like SMBB, as these types of things would be far too frustrating to play with and competitive play would be utter garbage. SMBB plays the game at the speed of your latency. Why you lag has little or nothing to do with Nintendo, and everything to do with the current state of the internet. You can certainly get some great games going, as long as you apply the information I've supplied here.

Getting the USB LAN adapter reduces latency.
This is actually a myth. Unless you have a lot of other appliances using the same frequency or your Wii is really far away from your wireless router, your ping won't be effected. Although, you will get slightly less packet loss using the adapter.

The 'Servers' are running horribly today!
Also a myth. I'm sure on a very very rare occasion, this might happen. However, since this service sets up a direct connection between you and your opponent, almost ALL latency issues arise because of networking issues.

What is the ideal latency for a game of SMBB?
To play Brawl competitively, you'd want the game to run flawlessly. In order to run the game flawlessly, you need to get a latency of 17ms or less. 1 MS (1 Millisecond), is 1/1000th of a second. 17 milliseconds is 17/1000ths of a second. If you divide 1000 by 17 you get a number close to 60, which would be indicative of how many frames per second the game would run at. Since the game naturally runs at 60 frames per second, you wouldn't want it to run any slower than this in a competitive environment. 30ms, though it sounds really fast, is almost half this speed (or 33 frames per second) !

-To play competitively, I would suggest no more than 20ms of delay.
-To play semi-competitively, I would suggest playing at no more than 30ms.

Every connection gets about 10ms of delay regardless of how far away the opponent is. With my figures of 5ms per 100 miles, that would indicate that you'd never want to play anyone more than 200 miles away for a competitive game and 400 miles away for a semi-competitive game.

EDIT: If you're like me, and you want the game to run great every time, get a program that can ping IP's and exchange IP's using http://whatsmyip.org/




I'll end this here. Hope this helps some of you, and I can finally link people this and save myself time explaining.

 

[LouisLeGros]

conclusion:
LAN FTW


you might even be able to take out the AN

 

[Tofu Beast]

why don't you keep your screen name the same?!

 

[KOdEx]

A lot of this is not entirely true I'll point out my major beefs.

My connection is super fast!
First, I'll dispel the myth of 'super fast' connections. With any connection there are two main factors: Latency and Bandwidth. Bandwidth is how much data you can transfer and is barely important in playing online games. As long as you don't download while playing and get more than 5k uploading and downloading, you're golden. Latency, on the other hand, is determined by server hops and the quality of material used by your network. Some ISP's are worse than others, but most broadband services are around the same quality. AT&T's DSL service is the only bad one in my area, and if you game, avoid this ISP like the plague.

Bandwidth is as important as latency in online gaming especially in fighting games like brawl because so much time sensitive information is traveling over the wire the more you can send and receive the better. Also there's no such thing as a 5k upload or download because thats not how data transmission is measured (usually, especially with data that small, 8 Megabits per second is equal to 1 Megabyte per second). It's measured in bits per second, and if you mean 5Kbps thats pretty much the equivalent to a 56Kbps modem, definitely not gaming on that. Need at least 768Kbps/128Kbps connection to even have a chance.(That's down/up speed).

Getting the USB LAN adapter reduces latency.
This is actually a myth. Unless you have a lot of other appliances using the same frequency or your Wii is really far away from your wireless router, there should be no lag problems.

Semi true, the best connection you could get out of your wii on a wireless connection is 54Mbps between the Wii and the router on a hard wired connection you can get 100Mbps effectively pretty much doubling the rate the communication rate between the devices(not incoming communication as that relies on how much bandwidth you pay for, unless you have a 55Mbps connection or greater (not likely though ) ) but for sending data you'll have a 100Mbps connection to your router which will then break it up into packets and send them to your opponent(s) at the upload speed you pay for. Using the wire also greatly reduces the chances of dropped packets which is the best reason to get it imo! Always use the wire if you can. Kinda defeats the purpose if both parties don't use it though at least you can say not my fault.

Don't let this happen to you because your wireless internet decided to take a break, this is what happens when the game compensates for dropped packets/lag:
http://news.filefront.com/halo-3-worst-betrayal-ever/

The 'Servers' are running horribly today!
Also a myth. I'm sure on a very very rare occasion, this might happen. However, since this service sets up a direct connection between you and your opponent, almost ALL latency issues arise because of networking issues.

Seems more like a misunderstanding of the word server kinda like there's a misunderstanding of Megabit(Mb) and Megabyte(MB).

Other than that Interesting breakdown of the ideal latency for brawl though it's like winning the lottery to get a 17ms response time.

 

[Gluttony]

I'm not sure your information regarding the lan adapter. I occasionally play friendly matches from North Carolina with someone I know from Canada. Randomly we'll get horrible ammounts of lag that renders the game unplayable. After these lag spikes the game would move smoothly only to do it again but ever since she got a the wired lan thing we've had hardly any lag at all. The game never becomes frustratingly slow.

 

[bnestman]

Seems like you had some good points. My matches always have horible lag and everytime i spectate a match it looks to be running smoothly...

 

[IWuvGeno]

A lot of this is not entirely true I'll point out my major beefs.

For everyone reading this, just ignore this statement. Apparently I struck a chord with this guy, and he decided to say something that he failed to support. I think only one of the statements he made is valid.

Bandwidth is as important as latency in online gaming especially in fighting games like brawl because so much time sensitive information is traveling over the wire the more you can send and receive the better. Also there's no such thing as a 5k upload or download because thats not how data transmission is measured. It's measured in bits per second, and if you mean 5Kbps thats pretty much the equivalent to a 56Kbps modem, definitely not gaming on that. Need at least 768Kbps/128Kbps connection to even have a chance.(That's down/up speed).

Bandwidth shouldn't even be considered if one isn't downloading or uploading at max speed and uses any form of broadband. What you said it so untrue it hurts. Brawl wouldn't send or receive more than 5-10 kilobytes/second. The number is a fixed amount as designated by the netcode, so it isn't better to send 'more' over the wire. Since the only data being sent are position synchs, velocity vectors, inputs, etc. the number is very small. Yes, I said kilobytes, because modern pingtests will convert bits into bytes. A kilobit and a kilobyte are only different by a factor of 8.

A 56k modem would only be bad because of latency (wow!) and a slightly slow upstream rate. Turn on your netgraph, and you'll apparently be in for a big shock on how little bandwidth your games consume. Latency is key.

Semi true, the best connection you could get out of your wii on a wireless connection is 54Mbps between the Wii and the router on a hard wired connection you can get 100Mbps effectively pretty much doubling the rate the communication rate between the devices(not incoming communication as that relies on how much bandwidth you pay for, unless you have a 55Mbps connection or greater (not likely though ) ) but for sending data you'll have a 100Mbps connection to your router which will then break it up into packets and send them to your opponent at the upload speed you pay for. Using the wire also greatly reduces the chances of dropped packets! Always use the wire if you can. Kinda defeats the purpose if both parties don't use it though at least you can say not my fault.

Again, you won't use even a small fraction of either of those transfer rates. But yes, you have a point about packet loss. The amount of packet loss is small, but one should consider it. I'll make that change, but it's pretty minor.

Seems more like a misunderstanding of the word server kinda like there's a misunderstanding of Megabit(Mb) and Megabyte(MB).

Other than that Interesting breakdown of the ideal latency for brawl though it's like winning the lottery to get a 17ms response time.

No one misunderstood the words Mb and MB here, but thank you for introducing the bit for anyone who needed the lesson. I have about 15 friends on my buddy list who get 20ms or less. I guess I should go pick some lucky numbers at 7-11.

I hope you learned your mistakes and don't go around spreading any of those myths about networking. There's enough confusion as it is.

 

[Zek]

Yeah, bandwidth means almost nothing to online games. There just isn't that much information to transfer. It's all about the latency.

Don't let this happen to you because your wireless internet decided to take a break, this is what happens when the game compensates for dropped packets/lag:
http://news.filefront.com/halo-3-worst-betrayal-ever/

Also, you're wrong about this too. The whole problem is that Brawl does not do anything like this. There's no prediction or lag compensation - if it isn't receiving info from one player, it will stop and wait for it. And its responsoe to high latency is to not perform your commands until after they've been processed, aka input lag. If Brawl used prediction like most online games, the input lag would be gone, but in exchange you'd have lots of bizarre stuff happening like being hit by attacks that you dodged or your attacks not doing any damage when they land.

 

[Monshou_no_Nazo]

The only real advantages I see in using a USB to Ethernet adapter is being able to bypass a router and plug directly into your modem to cut an ms or two of response time off (of course, I am willing to bet you can't actually do this if the Wii requires router specific settings), as well as not interfering with other wireless devices, but I wouldn't count on 100 Mbps or higher since your connection bottlenecks at the lowest speed, which tends to be your internet connection itself anyhow.

Also, you're wrong about this too. The whole problem is that Brawl does not do anything like this. There's no prediction or lag compensation - if it isn't receiving info from one player, it will stop and wait for it. And its responsoe to high latency is to not perform your commands until after they've been processed, aka input lag. If Brawl used prediction like most online games, the input lag would be gone, but in exchange you'd have lots of bizarre stuff happening like being hit by attacks that you dodged or your attacks not doing any damage when they land.

This guy cannot be more correct. Smash Bros' input lag is just a different way of handling lag. Compared to games like Gunz: The Duel where everyone complains about laggers they can't kill and that you have to practically shoot at invisible hitboxes and predict where someone is going to move and shoot right in front of them, in Smash Bros they make it so that hits actually land in exchange for the player having to time them differently. I actually prefer the other kind of lag compensation, but Smash Bros' way at least works for what it accomplishes.

 

[dumpsterdiver]

Why don't I lag as badly on other games?
Most internet games use certain tactics to reduce the necessity of super low latency (called prediction), unlike Super Smash Bros Brawl. As you play a game, if you've ever noticed a player suddenly teleporting backwards (rubber banding) or stopping and then speeding way up, you've noticed the results of prediction code. This type of coding wouldn't work for a game like SMBB, as these types of things would be far too frustrating to play with and competitive play would be utter garbage. SMBB plays the game at the speed of your latency. Why you lag has little or nothing to do with Nintendo, and everything to do with the current state of the internet. You can certainly get some great games going, as long as you apply the information I've supplied here.

Current state of the internet, are you serious? No, no, no no no no. While you're right about the "rubber banding," commonly seen in online FPS', the lack of such a technique is not game-breaking. There are many other games that cannot use that technique, such as other fighting games, many of which play great. The way I see it, the biggest problem is Nintendo, because they copped out and gave us a terrible P2P system. All players on a server, in this case, one room of smash, are connected to a host. The host runs the main game, and uploads information to all other players, including button input and any random elements. The players upload their button inputs to the host, and once all 4 pieces are together, a frame is complete. Input lag is caused by the host, or any player, not being able to keep up. So, by the time frame input 1 gets to the host, it should be frame 25 by now (an extreme case). These other 24 frames aren't lost, they just wait in queue. Theoretically, all players see the same frame at the same time, although I would think the host would see it a little early (where time difference is half the latency (since latency is a round trip), host latency to self is 0, guy A is 7ms, etc). However, when you're seeing frame x, you are inputting for frame x + 24, so you wont see that input come up for over a third of a second. A big problem with this system is that, for the most part, most people don't have connections that were designed for serious hosting. It would be far better if Nintendo had set up servers with rooms that 4 players could connect to, much like other online games (xbox live is an example of this, and I believe PSN). These servers would be able to handle the uploading more efficiently and reduce frame lag. Go play some online fighting games on those systems, like VF5, SF2:HF, etc, or , to a lesser extent, use the Kailera Network to play SSB64 on Project64, or MAME titles, generally these games work much better (and what's the isolated variable? The host).

That's possibility one, by the way, and it's the most likely scenario. Scenario B is that all players are "hosts" and upload data independently to each other player, rather than to a common local, but that would have a tendency towards desynchronization (ever had a replay not work right, input gets flubbed up, speed changes, etc? Imagine that in a game). I find this possibility really unlikely, it just isn't a logical way to do it. I don't even know why I mentioned it, I don't think any game uses it.

What is the ideal latency for a game of SMBB?
To play Brawl competitively, you'd want the game to run flawlessly. In order to run the game flawlessly, you need to get a latency of 17ms or less. 1 MS (1 Millisecond), is 1/1000th of a second. 17 milliseconds is 17/1000ths of a second. If you divide 1000 by 17 you get a number close to 60, which would be indicative of how many frames per second the game would run at. Since the game naturally runs at 60 frames per second, you wouldn't want it to run any slower than this in a competitive environment. 30ms, though it sounds really fast, is almost half this speed (or 33 frames per second) !

-To play competitively, I would suggest no more than 20ms of delay.
-To play semi-competitively, I would suggest playing at no more than 30ms.

The only real problem I have here is that those delays would not decrease framerate, unless the game got really backed up. The game is constantly sending packets, and the latency is how long one packet takes to make a round trip, not the time in between packets. So with any latency, the game is offset by half the latency (along with other factors, like processing time, packet construction and sequencing, etc.).

Every connection gets about 10ms of delay regardless of how far away the opponent is. With my figures of 5ms per 100 miles, that would indicate that you'd never want to play anyone more than 200 miles away for a competitive game and 400 miles away for a semi-competitive game.

That just isn't true, it's very possible to get a connection under 10ms latency. Also, where are you getting these figures? I think you made them up.

EDIT: If you're like me, and you want the game to run great every time, get a program that can ping IP's and exchange IP's using http://whatsmyip.org/

You mean the command prompt?

 

[The Executive]

You mean the command prompt?

*laugh* Funny, but with the command prompt there's much unnecessary scrolling of information and general ancientness. I prefer GUI-based apps.

 

[IWuvGeno]

Current state of the internet, are you serious? No, no, no no no no. While you're right about the "rubber banding," commonly seen in online FPS', the lack of such a technique is not game-breaking. There are many other games that cannot use that technique, such as other fighting games, many of which play great. The way I see it, the biggest problem is Nintendo, because they copped out and gave us a terrible P2P system. All players on a server, in this case, one room of smash, are connected to a host. The host runs the main game, and uploads information to all other players, including button input and any random elements. The players upload their button inputs to the host, and once all 4 pieces are together, a frame is complete. Input lag is caused by the host, or any player, not being able to keep up. So, by the time frame input 1 gets to the host, it should be frame 25 by now (an extreme case). These other 24 frames aren't lost, they just wait in queue. Theoretically, all players see the same frame at the same time, although I would think the host would see it a little early (where time difference is half the latency (since latency is a round trip), host latency to self is 0, guy A is 7ms, etc). However, when you're seeing frame x, you are inputting for frame x + 24, so you wont see that input come up for over a third of a second. A big problem with this system is that, for the most part, most people don't have connections that were designed for serious hosting. It would be far better if Nintendo had set up servers with rooms that 4 players could connect to, much like other online games (xbox live is an example of this, and I believe PSN). These servers would be able to handle the uploading more efficiently and reduce frame lag. Go play some online fighting games on those systems, like VF5, SF2:HF, etc, or , to a lesser extent, use the Kailera Network to play SSB64 on Project64, or MAME titles, generally these games work much better (and what's the isolated variable? The host).

That's possibility one, by the way, and it's the most likely scenario. Scenario B is that all players are "hosts" and upload data independently to each other player, rather than to a common local, but that would have a tendency towards desynchronization (ever had a replay not work right, input gets flubbed up, speed changes, etc? Imagine that in a game). I find this possibility really unlikely, it just isn't a logical way to do it. I don't even know why I mentioned it, I don't think any game uses it.

Other games you mentioned probably don't use physics engines, therefore requiring such a synchronous system isn't as necessary. Brawl must use a very complex system of double checking incoming data, with perhaps both position syncs as well as velocity vectors being transferred. This could also explain a little delay - or even dropping after too long. If the physics engine ever desynched... that would be disastrous.

Be sure to follow the link below to get educated. It's the state of the internet my friend, just do a little more reading. You're on the right track.

The only real problem I have here is that those delays would not decrease framerate, unless the game got really backed up. The game is constantly sending packets, and the latency is how long one packet takes to make a round trip, not the time in between packets. So with any latency, the game is offset by half the latency (along with other factors, like processing time, packet construction and sequencing, etc.).

My question is: if delays don't decrease frame rates, then how would you suppose they fill in the rest of the time? A possibility is that Brawl will start to skip frames when the latency is high enough (to create the illusion of the game running faster). It sounds like you need a lesson in online gaming! Read 4.1:
http://allisbrawl.com/forum/topic.aspx?pid=34487#p34487

That just isn't true, it's very possible to get a connection under 10ms latency. Also, where are you getting these figures? I think you made them up.

I've been running ping tests for a long time now, between people all over the US. I definitely didn't make them up. I've been measuring distances and between about 100 or so test routes. So, it isn't the final say in the matter, but it's fairly reliable. And you aren't likely get under 10 ms to anywhere, no. I'm pretty sure you just make stuff up trying to debunk people.

You mean the command prompt?

You can use command prompt if you like, although I wouldn't recommend it personally unless you don't know of any software alternatives. Or you like wasting time.

 

[dumpsterdiver]

Other games you mentioned probably don't use physics engines, therefore requiring such a synchronous system isn't as necessary. Brawl must use a very complex system of double checking incoming data, with perhaps both position syncs as well as velocity vectors being transferred. This could also explain a little delay - or even dropping after too long. If the physics engine ever desynched... that would be disastrous.

Be sure to follow the link below to get educated. It's the state of the internet my friend, just do a little more reading. You're on the right track.

Are you serious? At first I thought you had a few good ideas and maybe knew what you were talking about, but seriously? Of course other games have physics engines, that's the stupidest thing you've said so far. Brawl's is far from intricate, as well. Besides, I also mentioned playing SSB64 online, and it works great. Brawl and 64 should behave similarly online. Also, get your head out of the clouds, you don't know everything. Seriously, state of the internet? NO. Just no. Also, if you're transferring position 60 times a second you don't need a velocity vector. A velocity vector would allow the game to determine a path on it's own, which isn't needed if you know the position, it isn't even a good idea. Rather, it needs position and state (the animation a character/object is making, including orientation). Double checking data could be a cause for delay, though, but that would just be bad programming. You'd only need a velocity vector if your own system was filling in frames, which doesn't seem to be the case (framerate drops occur and slow down the game, not just make it choppy, meaning it's still receiving frame data 60 times a second). Velocity could be needed if the host wasn't sending 60 fps, but I don't really have a way to test this.

My question is: if delays don't decrease frame rates, then how would you suppose they fill in the rest of the time? A possibility is that Brawl will start to skip frames when the latency is high enough (to create the illusion of the game running faster). It sounds like you need a lesson in online gaming! Read 4.1:
http://allisbrawl.com/forum/topic.aspx?pid=34487#p34487

I already answered this. The time is filled by sending more packets. You still receive 60 frames worth of data every second, but the information arrives late, hence the delay. If the delay becomes too great, the host (or even the clients) may be unable to keep up, in which case a frame drop would be required, since data can't be processed and sent fast enough. The thing is, though, Brawl doesn't simply skip frames. If you've ever played Brawl online, you've probably also hit huge fps drops, where the game actually slows down. It still tries to play every frame, even if they're coming in too slow. This is known as lazy programming, as framerate drops are easier to work with than frame skipping.

What's more, I read 4.1 as you suggested. Also riddled with lies. Latency does NOT determine how many packets you receive per second. The server will not wait for a response to send out another packet, it just does it. What determines the number of packets per second is the server's tickrate, which is the number of commands sent to all players each second. For example, a 100 tick server runs the game at 100 fps. That means every .01 seconds it receives input, calculates one step (plays one frame of the game), and sends out whatever information it needs to each player. Latency determines how quickly each packet gets to its destination, but latency is round trip, so subtract processing time of the ping due to the use of ICMP (I'll get there in a second) and cut it in half. Since we're playing Brawl, we want a 60 tick server, but each "tick" will arrive to the clients a little late due to latency.

I've been running ping tests for a long time now, between people all over the US. I definitely didn't make them up. I've been measuring distances and between about 100 or so test routes. So, it isn't the final say in the matter, but it's fairly reliable. And you aren't likely get under 10 ms to anywhere, no. I'm pretty sure you just make stuff up trying to debunk people.

I'm going to smack you for insisting this. Ever played Counter-Strike? Most people with a decent connection can bring in a ping of 7 or less (I've played at 4 before) at a nearby server. There is a problem with your test: pinging a server uses ICMP, a slow *** protocol that is mostly used for pinging, and little else (error messages over networks, too, but who cares). It gets the job done and works as a relative speed test, but its low priority makes it useless for online gaming. [Most] online games use UDP to ping a server (really, the server pings you and then sends you the number, at a rate it determines) as well as for data transfer, which is much faster. ICMP requires an echo, sending back the same packet. With UDP, a ping request can be tacked on to normal game packets. UDP also isn't checked, things can go out of order or missing. THIS IS ADVANTAGEOUS TO GAMING. It is better for a packet to be lost than to simply WAIT for it, as that will slow everyone down. This makes me wonder if the packets are being checked (TCP maybe?) and the developers decided it would be better for it to run slow for everyone instead of one person occasionally missing a frame (in other words another bad decision on Brawl's part).

You can use command prompt if you like, although I wouldn't recommend it personally unless you don't know of any software alternatives. Or you like wasting time.

If anything, searching for a program to ping a server when you already have that ability is a waste of time. It's not even hard to use, it's just "ping xxx.xxx.xxx.xxx" That's far easier than booting up an application every time you want to ping something. But I'm no way suggesting that you're a scrub who's afraid of a simple command prompt.

The point of all this text is simple: you are wrong about things. MOST IMPORTANTLY: the lag is NOT the fault of "the state of the internet," its a crappy online system. Other systems work (even with similar games), Brawl doesn't. It doesn't take a genius to realize the problem is with the **** game. Please stop spreading your filthy lies (perhaps lies is a strong word, but you have clearly been misled by SOMETHING). I know the desire to be an internet hero (all hail that guy who told me about my game's lag!), but PLEASE read up on this stuff first. It doesn't take much, you could probably get most of this stuff off of Wikipedia if you wanted.

 

[Randofu]

To be honest, you're both right.

Latency is critical. Given Brawl's system as it is, there's not much that can be done other than to try to play against people with whom you have low latencies.

Brawl's system could be improved somewhat (but probably not much) by placing a ridiculously large number of servers throughout the network to act as hosts. This would be very expensive; Microsoft can easily throw their weight into getting such a system deployed (they probably have such things for other purposes already), but Nintendo is a humble video game company. You can't expect them to dish out tens of thousands of dollars to slightly improve the online game play for only a small fraction of the users. The Nintendo online play problem is not a technical one; it is a business problem. Currently, it is not profitable for Nintendo to expand their online system.

I'm considering coming up with a Java "friend finder" application that would essentially let you find other people using it and profile what your connection to them would look like. I think it could be a really cool piece of software, so my only question is, would people really be willing to use it? Basically I'm imagining something that, to the user, would work like an instant messaging client (you just leave it on all the time pretty much), and whenever you go in to look at it, it recommends to you the people with whom you'll have the best connections (in terms of latency, jitter, and loss).

Would people use this?

 

[dumpsterdiver]

To be honest, you're both right.

Latency is critical. Given Brawl's system as it is, there's not much that can be done other than to try to play against people with whom you have low latencies.

Brawl's system could be improved somewhat (but probably not much) by placing a ridiculously large number of servers throughout the network to act as hosts. This would be very expensive; Microsoft can easily throw their weight into getting such a system deployed (they probably have such things for other purposes already), but Nintendo is a humble video game company. You can't expect them to dish out tens of thousands of dollars to slightly improve the online game play for only a small fraction of the users. The Nintendo online play problem is not a technical one; it is a business problem. Currently, it is not profitable for Nintendo to expand their online system.

I'm considering coming up with a Java "friend finder" application that would essentially let you find other people using it and profile what your connection to them would look like. I think it could be a really cool piece of software, so my only question is, would people really be willing to use it? Basically I'm imagining something that, to the user, would work like an instant messaging client (you just leave it on all the time pretty much), and whenever you go in to look at it, it recommends to you the people with whom you'll have the best connections (in terms of latency, jitter, and loss).

Would people use this?

Setting up servers would cost a bit more than that, but yes. However, Nintendo is NOT a humble little video game company, it's a monster. They could totally afford it, but it'd be pointless because their entire concept of online gaming is a joke (not just Brawl, either... just everything). If they got over their irrational fear of letting us communicate and put up servers, they could even CHARGE us for online service ala XBOX live. It just wont happen.

As for your idea, it seems nice but I doubt anyone would really use it. It seems a bit cumbersome, especially since many people already keep AIM, Skype, MSN, etc. open at all times anyway, and without being able to connect it to the Wii key features like seeing who's online at the moment are gone. Also, it would cost you money to host the server.

However, if you made a plugin for say, Pidgin (open source AIM client), that can automatically ping your friends, that might be nice. Or a combination AIM client/autopinger (for lack of a better word) plugin for Firefox. An online app could work, too, but that would also require you to host the service.

I don't mean to rain on your idea, but the benefit to the few who would install it might be outweighed by the costs. If you use a preexisting system like AIM, then there's no need to host a server to hold everyone's user names and connect everyone.

 

[IWuvGeno]

LOL. Dumpsterdiver. Apparently you don't know why the counter-strike rates make your ping display wrong, it seems you've never programmed a physics engine, have no idea how they work online (or the lack thereof), and have no idea how packets are sent.

I'll give you the answers tomorrow, as I don't have time to write a long post tonight. For now you can anxiously await.

 

[KOdEx]

For everyone reading this, just ignore this statement. Apparently I struck a chord with this guy, and he decided to say something that he failed to support. I think only one of the statements he made is valid.

I actually enjoyed your post, it was very interesting. Your making assumptions thank you.

Bandwidth shouldn't even be considered if one isn't downloading or uploading at max speed and uses any form of broadband. What you said it so untrue it hurts. Brawl wouldn't send or receive more than 5-10 kilobytes/second. The number is a fixed amount as designated by the netcode, so it isn't better to send 'more' over the wire. Since the only data being sent are position synchs, velocity vectors, inputs, etc. the number is very small. Yes, I said kilobytes, because modern pingtests will convert bits into bytes. A kilobit and a kilobyte are only different by a factor of 8.

Don't get me wrong i never said ping didn't matter ping is still king, which is why you can't play on a satellite connection lol even though they get good bandwidth. Just having extra bandwidth helps. Meaning you can never have to much of it. I'd like to see this netcode you have it's very interesting.

Again, you won't use even a small fraction of either of those transfer rates. But yes, you have a point about packet loss. The amount of packet loss is small, but one should consider it. I'll make that change, but it's pretty minor.

Minor if you have a good wireless connection but it still a possibility even if you do.

I hope you learned your mistakes and don't go around spreading any of those myths about networking. There's enough confusion as it is.

The only mistake I've made is maybe I wasn't clear enough in my explanations I suppose. Sorry.

And to the video posted, it's just an example of what lag can do in a game that compensates for it. I know in Brawl it'll just stop working if it's bad enough because it's happened to me. :D

 

[Randofu]

If I made such a thing, it would be a distributed P2P system, so the only thing the "server" would have to do is bootstrap it (which I could do from my home computer).

Heck, I could probably pretend that it's part of my research (I'm a networking grad student) in some way, and get my university to host it. I don't think it would really be anything NEW though, so I might have a hard time justifying it.

But you're probably right anyway. I doubt people would use it.