Lag Tactics


His throws are pretty able for abuse, but it isn’t a big deal compared to other characters and their lag abusing abilities.


so what’s your excuse when a match is lag free? You still lose.


That’s not what he meant. he didn’t say you should not post, but you should not create NEW threads for general stuff (I made that mistake too when I first joined).

There are character specific threads if you want to discuss stuff related to a character

And then there is a HDR general discussion thread and a marsgodtier thread where you can discuss anything that is not related to mars.

Yeah boxer’s dash punch and whiff dash into grab are hard to react. So is ryu lk tatsu into throw. In general throw setup is harder to counter online.

I once have a friend using o ken and just keep repeat jab shoryu, and my job is to sweep him, it’s hard…


sorry i just meant it in a “here’s some free advice” kind of way. i’m in no way a mod or anything


Hardware, software, and network performance are improving exponentially over time.

Relatively soon, lag will not be an issue and this will profoundly affect the fighting game scene. The online experience will come to match today’s offline experience, and in fact will probably be preferred for a number of reasons.

There is a lot of idolatry today for the “one true” experience of offline play, and it’s amazing to me that the strongest adherents of the cult of offline are seemingly so blind to the truck that is about to hit them square between the eyes.


I have trouble w/ Guile/DJ sonic booms…


This is kind of off-topic, but okay, I’ll take this bait. Let’s take the four claims here.

  • Hardware performance is improving exponentially over time.

This is probably the truest of the four, see Moore’s law and all. But it’s still pretty questionable. Think back to the 1990s: you could buy a state-of-the-art computer, and it would be completely obsoleted within three or four years. You’d buy a 150 MHz Pentium Pro in Nov 1995 and replace it with a 450 MHz Pentium II in Jan 1998, because your 1995 computer simply couldn’t run anything current anymore. But chances are pretty good that if you bought a computer in 2005, it still runs basically everything you want in 2010, and if you look at the CPU clock speed it should be pretty close to 2.5 GHz, both in 2005 and in 2010.

Why? A big reason for this is that our hardware is already getting really close to the limits of physics. We’re already making things so small that quantum effects happen between parallel circuit lines and screw things up. We have extreme trouble making clock speeds faster than about 3.5 GHz, because of fundamental limits of thermodynamics. And the absolute transistor size limit of a single molecule’s size is already pretty damn close.

Sure, hardware performance has improved in the past 5 years, but it’s nothing compared to the breakneck speed of the 1990s - if anything, the hardware market is getting way more stable over time, it’s not exponentially accelerating. And I find it reasonable to believe that the next ten years will also be ten years of greater stability, not exponential growth.

  • Software performance is improving exponentially over time

Yeah okay, I have no clue where you got that idea. Software is hella hard, and I’d be surprised if we improved more than linearly over the past 20 years. We coded console games in C++ 15 years ago and we still do today, and while yes the overall tools and processes have improved, it’s been closer to logarithmic growth than exponential growth, heh.

  • Network performance is improving exponentially over time

Really? Because it sure seems like my home bandwidth has been pretty damn stable for the past 10 years.

Actually, there’s two components of performance here: bandwidth and latency. I have no real problem believing that bandwidth will improve in the near future, you can just install fat optic cables everywhere, and bam you’re done. Though that process isn’t anywhere near exponential growth, heh.

But the bigger concern for us is latency - that sure isn’t going to improve exponentially. And the reason is that it’s goddamn hard to beat the speed of light. The Earth’s circumference is about 40000 km, and light travels at about 300000 km/s. If you want to send information to the other side of the globe and back, even if you had the theoretical perfect system where you instantly send and recieve pure photons in a perfect curve hugging the planet, without any relay systems or intermediaries at all, all within a perfect vacuum, it would still take 133 milliseconds.

And in the real world, we’re dealing with electrical pulses transmitted over copper wires, that traverse a whole bunch of routers, relay stations, converters, etc. I do have faith in the power of engineering, but we’re just not gonna beat the speed of light anytime soon.

So that brings me to the last claim:

  • Relatively soon, lag will not be an issue and this will profoundly affect the fighting game scene.

Because of what I’ve just explained, especially w.r.t. latency, I think it’s safe to claim that lag will continue to be a very real issue. Sure, clever engineering tricks might reduce ping times by, like, 20% in the next 10 years maybe? I just pulled that number out of my ass, but considering the absolute limit of the speed of light (and the softer limits of hardware, software, and network technology), I’d be hella surprised if we even got there. My point is, even that hard-earned 20% isn’t going to change much. We’re simply not looking at a future where “lag will not be an issue”.


The speed of light is a limiting factor and I should have mentioned that in my post. Software is hard, that is also true. But it seems like even though you made a rather long post, your criticism is largely built around my use of the word ‘exponential’. Fair enough, it was a generalization.

Let me restate: the presence of lag will diminish to the degree that we perceive very little or even no difference in offline and online play. Lag only has to reduce to the level where we cannot perceive it.

The past is littered with people and companies who could not understand (or would not admit) the dramatic changes that the future holds, even the near future.

To be honest, your position reminds me a little of the disk manufacturers described here:

The Innovator’s Dilemma


10 years ago, the latency between Brazil and US East cost was around 90 ms in average. It would often return 80 ms, with spikes being around 140 ms. Now the latency is around 180 ms, with spikes around 5 seconds. It only got worse, supposedly because they are using different standards right now, so the protocols need to be changed in between the end points.

It is not possible to make lag such as 60 ms unnoticeable in SF2 or FPS. It is just not possible, in fact. And latency smaller than that around the globe will never happen, exactly due to the light speed limit.


Your experience with lag in between Brazil and the US is not generalizable.

Also, I’d say that the use case where the two players are on the opposite sides of the planet is not the normal use case.


I doubt this will happen in the near future for fighting games. If it does, that’s fantastic. I don’t think anybody would be disappointed by that, so I’m honestly not sure where the conflict is, but generally I just don’t see this happening for a very long time, aside from playing against people within a very close range.


Fair enough, we’ll agree to disagree :slight_smile:


The distance from Sydney Australia, to New York, New York, USA is about 16,000 km, the speed of light is 300,000 km/s so there’s no way to get better than 1/20 of a second (3 frames) unless you tunnel through the earth.

New York to Los Angeles is about 4,000 km - or about 1/80 of a light second (3/4 of a frame). I’d be rather surprised if people could get less than a frame of lag over that distance. I doubt it’s possible with any consumer grade service.


I already said that I think the use case where the two players are on the opposite sides of the planet is not the normal use case.


It is possible that some sort of wireless technology could be developed that could penetrate the planet, so the maximum distance traveled would be the diameter instead of half the circumference. 12,742km/300,000km/s = ~0.042 seconds, so that is a slight improvement.

Also, while input lag netcode is generally bad, the main reason is that it isn’t consistent with offline input lag. If you get the base input lag down as low as possible, then add artificial input lag to the offline, you could use that difference to smooth out online play without any differences. PSN HDR apparently has 7 frames of input lag offline and doesn’t feel too bad. I don’t know about console, but on PC, it should be possible to get down to 1 frame of input lag, so you can add 6 frames to keep it the same. 6 frames is 1/10th of a second, so you could get a 100 millisecond ping to be exactly the same as offline. Rollback netcode would exist for anything above that, and generally rollbacks less than 100 milliseconds aren’t too bad.

But, yeah, even with those two things, it won’t be enough to make worldwide netcode the same as offline. Maybe we’ll get some quantum entanglement network infrastructure eventually?


no thanks


I didn’t have any plan to revisit this thread, but just saw this:

Progress in Algorithms Beats Moore?s Law Algorithmic Game-Theory/Economics


Let’s wait 10 years to find out who is right :wgrin:


LOL, yeah, that’s pretty much it.