It goes brings lot of things to my mind though.

For example, let's say I hooked up two NESes to play smb using 1 controller output to two TVs... one uses romhack and one uses the original cart. Could I do this to know subpixel values that tells me every specific of whether I executed things correctly?

How about if I made a device that just moves my fingers to press the inputs like a TAS? :D

PS: If you tape down the "B" button, you can't press "start" on the title screen to start the game.

Personally I would say no, since you're using a physical object to assist you. He does have a small point about using splits though. In the past, I also considered taping down the B button, but never did.

Technically it's controller modification I'd say. I think I would reject a run that had it.

As for splits, sure I guess they help, but what about chat on twitch even then? Like technically, chat can help quite a bit too (let's say they could identify framerules by the coin flash: me). It gets into semantics a little bit.

I just wouldn't classify splits as a "tool" necessarily when there's also a way to identify the same thing in-game,

I've been doing 1-2G attempts a bit more now that I'm working on any% again. Once you've clipped through and need to jump back (move left) to the world 4 pipe, you typically need to do a frame perfect jump in order to prevent Mario from facing leftward.

If you hold left+down while making the turnaround, it forces Mario to face rightward and gives you some leeway on the jump to the left. The only issue might be that it might not be fast enough to save the frame rule if you don't jump frame perfectly. However, I believe it should work and should help slightly with consistency.

The same applies for the walljump in 8-4 by holding right+down while walljumping backwards in order to prevent an accidental walljump turnaround jump.

Just curious, does anyone know exactly why it works on PAL and not NTSC? My theory/guess would be that Mario can jump lower on PAL, but I'm not exactly sure that this is the reason.

I like "Any%" + "all levels" + "andrewg%"

Seems fair. Allow everything for any% and andrewg%, and make "all levels" just by banning the vine glitch.

I think both should be allowed.

I can see people not wanting suicide code because it's a code though. I'd definitely like to hear others' input.

As for continues, I guess I'm in favor of it. I find it amusing because that strategy results in a softlock (so you'd have to reset anyway ;)). I'm amused and happy that this is a thing at least. Lost Levels SNES any% has a similar kind of gimmick, so I'm OK with allowing it.

HOWEVER, this does require use of suicide code. I'm curious people's thoughts on using the suicide code in order to quickly kill yourself (restart any area with full health). Or in the above circumstance. Thoughts?

It's kinda like if you allow one, you have to allow both here.

That's neat! Nice work.

A few years back I made this post and recall it happening:

"I'm fairly sure that I didn't imagine this. Back 2 or 3 years ago when I was playing Pro Wrestling, I believe I somehow skipped the number of opponents I faced overall. I swear I only faced 12 opponents when I finished the game. Several things this could be: Using King Slender - it should add 5 matches to the total number of matches, however, I believe that losing in a specific manner may somehow screw this up. I have multiple copies of Pro Wrestling as well, it may have been some sort of version difference. It's at least worth experimenting on, but it may have been some strange fluke?"

However, I have yet to ever duplicate or see this happen again. Thought I would mention it here in case anyone wanted to look into it.

It could have just been my imagination, but I remember thinking it didn't add up right.

I'd like to see this implemented across all games. This was actually my suggestion for Twin Galaxies a few months back, but they're idiotic with how they do everything all the time.

Anyway, I like the idea because it contributes more submissions to the lesser leaderboards on games. I wish it were automated because I think people would really like the idea. Updating it for everyone doesn't seem entirely feasible with the amount of submissions we get, but it might be cool to see like the "monthly leader" at the end of each month or something.

Honestly, I always wanted this to be a thing with the exact method used. I think it's a pretty good way to measure ability across all of a games' categories.

Maybe we could create a program to automate the process. I'm not sure everyone would like this kind of leaderboard, but it would be pretty neat to have a leaderboard "master list" for each game on speedrun.com using just the main categories.

I find it strange that the rules:

1. Start timing at "start" press.

2. have a time adjustment rule to U.S. version

3. I just find it weird that we have a different standard between the timing scheme of SNES SMB1 and SNES Lost Levels. I don't really understand why this is the case.

4. I'm not so much against this, but is a little atypical. If a small difference between the versions is found later on, it could potentially be a bad thing. For example, with Super Mario World, the J version had a harder orb method that could save time over U.S. The only time difference between the versions was a single text display was only 31 frames. The standard with speedruns these days has just been to use the fastest version to avoid these kind of situations.

Just thought I'd mention it. I personally prefer starting from character control, but I guess it doesn't matter.

Hey Scott! Great to see you post, and I hope you're doing well. Where's your Bionic Commando avatar!? The timing you're referring to is correct. TG started timing from "start" from your 5:17 onward.

I felt like digging back through history a bit. I like to give credit where it's due, and so I wrote the names of a few key people and speedrunners who achieved great times way back in my SDA descriptions. Here's what I found:

https://web.archive.org/web/20060319063339/http://www.re-retro.com/2006/02/26/mario-race-hold-the-karts/

This video was posted in 2006, but this race was done sometime in 2002. They are surprisingly good for when this took place. I'm not sure their exact times, but I would guess these are oldest recorded Super Mario Bros. speedruns available.

Wow, more digging and some cross-referencing and I'm realizing that Jason Harmer's old Super Mario Bros 5:09 claimed time was actually recorded. Which he said he achieved sometime in 2006. I would use the race video as a date reference... 4/3/06

So Jason Harmer got a ~5:12 in Super Mario Bros. before 4/3/06 (if you're going off his claim from the interview about the super mario bros race):

Would you reset after each hard track? Would this be individual times added up? Might be fun.

I play games for world record most times because it is a fun challenge, but I don't expect WRs even on games I'm good at. It's always a pleasant surprise when it happens.

You need to believe you can get WR to do it and aim for that. However, you need to simultaneously realize that it takes a lot of effort and a ton of time, and that shouldn't be the sole reason for speedrunning a game. Being part of the community is much more fulfilling than getting world records.

Your final point was what I was about to say. As good as any run is, every run is improvable (with few exceptions). Even when 4:55 happens in SMB, there will always still be frames to save. No category can really ever be dead.

I also believe most can accomplish a wr with enough effort, but that might not be your goal. If it is, it can be a long journey, and is not always possible.

But... Practice makes perfect.

Maybe I should make a good tutorial to explain what I do know. This is I think how it works:

1. The longer the air time, the faster the bike speed
2. The faster the bounce, the faster the bike speed

Long air times are gained by switching down a lane while going off a triangular (speed) ramp. This lengthens the distance you can cover. Switching up a land while going off a triangular jump shortens the length of your jump, which is not practical for speed but for positioning.

The faster your "left" press when bouncing after going off a speed jump, the faster your bike will travel.

There are definitely things I still don't know about the game though. I know holding forward gains time off ramps, but there can be ways to gain small amounts of time (maybe it's all about the bounces?). I've always felt that the ideal method is to be on the ground as long as possible and always hold forward when going off ramps... I think ideally, you would want to do perfect lands (holding forward) instead of bouncing each time. Hard to explain everything I mean properly without using video or pictures.

Basically:

• I think you want to lean the bike all the way forward every time you go off a ramp.
• After going off a ramp, land on the ground without bouncing if possible (leaned all the way forward)
• Switch down a lane while going off the triangular (speed) ramps for maximum distance (air time and speed).
• When bouncing after going off the triangular ramps, press "left" for as little time as possible when landing and continue to do so from that point onward.

And to make matters worse, you scheduled it on my birthday. ;)

We still need to finish the one from 3 years ago :D

I won't be able to participate. Well, there's a slim chance.

I think track 2 would be the only track where we could realistically get "perfect" times in a sense. However, if you could find a way to consistently save frames, that would of course be useful knowledge. I would look into the optimal time to press "B," because I'm pretty sure that is a time savings we could use.

I think as far as the RNG, we're probably out of luck trying to make it consistent. However, I'd love you to prove me wrong.

Depending on when you press start can affect the initial acceleration which causes slight differences in finishing times. As far as I know, you can gain 2-3 frames with a lucky start. I also think there is a way to accelerate in a more efficient manner than just holding "B" the whole time (which gains a few frames). I don't think there's any real way to control the starting difference to always have good RNG, and I never tried better acceleration by pressing "B" at a very specific point because it messed up the "TEMP" gauge in certain tracks. However, I think tracks 1, 2, and 5 could use this for a slight time save... I don't know what the exact start method is that gains time, but it's minor.

I essentially ignored both of these factors since the times have never been like other games with individual levels in the fact that huge time gains can be made even on pretty optimal times with better bouncing and whatnot. Potentially, the times will reach a point where two runners have times milliseconds apart ( ;) ) that require good RNG, but I'm not sure that will realistically happen because of how much the speed varies and how much it affects the overall times.

I think the reason for this is because the game only registers inputs every 3 frames (if I remember right) and so depending on when you press start, the first "B" press that actually moves the bike might fall .04 seconds into the race. I think there might be a little more to it as to how it affects the outcomes in the races. There are some Excitebike TASers who could explain it more, because I don't fully. However, I've always felt the savings was too minimal to bother finding out. And if I ever got to a point where I wasn't improving my times, I would look into it :D

Glad you're researching it though, I would be interested in knowing exactly what's up. I haven't tested things for this game in years and years.