Saturday, January 2, 2010

VVVVVV, speedruns, and being the best (bonus section included)

(This article includes a special bonus section at the bottom!)

Recently released was the beta for a flash-based game called VVVVVV.  That's 6 V's.  You can learn a bit more about it at  Its designer is a gentleman by the name of Terry Cavanagh.

The final version will be released January 10th and if it stays like the beta, then I can assure you this game is unbelievably terrific in many, many ways.

Two related factors; one, it's very hard, and the music is terrific.  Why are these related?  Because you're going to die a lot and, consequently, you'll be listening to the same tracks looping over and over again, so the music HAS to be good.  And it is.

The gameplay is simple.  You can move left or right, and you can flip gravity.  Hit V, and you flip up to the ceiling.  Hit it again, and you start falling back down.  You can't flip until you've hit a surface, so you can't just mash V and hover in place.  You travel around a large world going from room to room, collecting shiny trinkets and rescuing your shipmates.  It's old-school in presentation, style, and--of course--difficulty.  This includes going into rooms that kill you unless you already know what is in them.  I typically find this to be a design flaw, but you repeatedly run into little respawn checkpoints so you almost never lose any substantial progress from dying, and the game demands precise platforming skills so even when you know where things are you tend to die anyhow.  It's just damn hard.  I liked this game way more than I think I should have.  I liked it so much, in fact, that I'm doing a speedrun for it, even though it isn't technically finished yet.

This is where the game's magnificent design becomes apparent to me.  Almost every room seems capable of being cleared at full speed, provided you have frame perfect timing, you understand the physics engine, and you know where everything is.  Some rooms you clear at ALMOST full speed by knowing just where to slow down and how to enter each room.  The game is basically a giant puzzle waiting to be solved.  It's a lot of fun, I recommend you check it out if you've got the $15 dollars to do so once it's released.

So this brings us to the second point; how you go about creating a speedrun.  This game is rather easy on you as far as designing the speedrun goes.  You never get any powers and you can't skip over levels because--in order to do a full 100 percent speedrun, collecting the trinkets scattered throughout--you simply have to go through the stages as normal.  In a Megaman game, for instance, you have to pick a path for beating all the robot masters so that you can use the powers to defeat the masters and zip yourself through stages as quickly as possible.  The order you do things in matters.

Not so in VVVVVV.  You don't get any abilities, and rescuing one crewmate won't really affect how you rescue the others (with one tiny exception that will save you one or two seconds overall).  So figuring out your path isn't that difficult.

Figuring out each room is where the real fun is.  I went through the first dangerous room somewhat slowly.  Over one enemy, under another, and then wait for the last one to drop so I could go over it.  After a bit of experimenting, I found you can--with perfect timing--jump UNDER the last enemy on its way up.  So this speeds up the room by about 1/2 a second.

Not much, really.  But with hundreds of rooms in the game, those halves would add up.  And if I didn't do each level as fast as I knew how to do, somebody might imitate my run EXCEPT for this one room, and beat me by a second and take my record.  If you want to have the best speedrun, you have to think like that.  Somebody just might care a tiny bit more than you, and practice a little bit harder, and get that half-second.

For instance, I remember reading on a board dedicated to Melee HRC and BTT a thread somebody made saying he had the new world record for the Game and Watch HRC.  He had topped the other record by something miniscule, like .2 feet.  All he did--I'm guessing--is inch the bag forward slightly while doing his bat-drop combos, and otherwise mimic the world record exactly.  But you know what?  He had the new record.  .01 seconds can mean the difference between a gold and silver in the Olympics.  I'm pretty damn sure an Olympic level athlete would NOT say "one hundredth of a second?  Eh, whatever, I didn't want the gold that badly.  What's the difference, really?"

The difference between winning and losing a close Smash match can be one more hit than the other guy.  It can be hitting each other at the same time, but you were a little closer to the blast zone than he was.  It can happen in the finals of a tournament.  Does it matter?  Yes.  Absolutely.  It will not be a trivial difference when it's the difference between having a record--or victory--and not having that record or victory.  So yes, mastering this stupidly hard jump to save 1/2 a second really does matter.

Then I learned that you can go through this first room with the right timing so you reach the third enemy fast enough to go over him without waiting at all, and you don't have to jump under him either.  I didn't know you could do that.  It's actually *easier*, and saves another 1/4 of a second.  That's a bit of a relief.

Next room leads you to your first trinket.  There's a checkpoint in here, so when you trigger it, if you die you will return to that checkpoint.  Trigger the checkpoint, drop into the room with the trinket, and then... there's another checkpoint and a lot of spikes.  Reach the trinket, leave the room.  Right?

Wrong.  If you do the room just right, you skip the checkpoint.  Get the trinket, kill yourself, warp one room back, save a second or two.  And so on.

Just about EVERY room in this game has a trick like this.  If you time a jump right, enter the room at the right angle, go as FAST as possible, you can blitz through this game.  And kill yourself a lot in the process because the timing is NOT forgiving.  Hell, the game's bad enough to do normally, let alone charging through at full speed.

But if you want to speedrun, there are a few basic tenets.  One is that it doesn't matter if something is hard.  If it's faster, you must attempt it, because somebody else will.  Second is that there is, very often, a faster way than what you have, and it's waiting for you to discover it.  You have to be willing to question what you believe is possible.  "I can't make this jump."  Actually, maybe you can, and you're just not doing it right.  And almost all the time, your movements can be just a little sharper, you can cut each corner just a little more closely or jump just a little earlier, and save yourself that tiny bit of time.  And through an entire game, it all adds up.

However... bear in mind that in a single segment run--a run where you start the game and beat it in one sitting with no saves--you are probably going to mess up, especially on very difficult games, and especially towards the end.  My record for beating this game is currently clocked at 19:14 by the game's timer.  The first five minutes or so, I won't tolerate anything but superficial errors.  By the end, however, I will begin to accept mistakes, because otherwise I'm wasting the run and the opportunity to practice other parts of the game.

As the run goes on, I will tolerate bigger and bigger errors because the third tenet is accepting your human limits.  I am GOING to make an error in the time it takes to complete this game, especially when I attempt frame perfect jumps after a long period of nonstop focus.  And if I'm recording, it may still be the best speedrun I've done, even with a few dumb mistakes in it.  It may be the best anybody's done, and I'm not going to sneeze at that.

And nobody says I can't replace my own run!  So I can choose to do an "inferior" run this time, skipping over a few of the harder tricks in favor of safer, more consistent methods, particularly towards the end when I don't want to waste an otherwise good run.  I can keep that as my current record, then I'll have something good to beat.  And if nobody else does it in the meantime, I hold the record.

If you've played Mario Kart and done the time attacks (or saw the Speed Racer movie :P), you know that when you try to do a time-trial on a given level, the game races you against the "ghost" of that previous record.  When speedrunning, I feel like I'm racing against the ghost of a person who does my run just a little better than me.  I make a mistake, and I imagine that some invisible competitor has, until that moment, done everything exactly like me... except without that mistake.  So I'm losing, and I have to step my game up.

Because this is not a blog about Speed Racer, Mario Kart, or VVVVVV, and it is a Smash blog, I should be talking about how this relates to Smash.  But I think the parallels should be obvious.  If you don't dedicate yourself to mastery, somebody with a little more dedication will do what you do, but better, and you will lose.  You can decide NOT to practice your Fox ditto chaingrabs, but then you will play somebody who goes pretty even with you, except for those chaingrabs, which he does better.  And that could be what clinches the match.  Little things add up.  Somebody who is just a little sharper, or faster, or a little more consistent will beat you out.

Then again, nobody is saying you have to be the best at anything.  That's a personal goal that you set for yourself.  But if you want to be better, you can't ignore things and say "it's too hard" or "it's not worth it."  Obviously you want to pick things that give you the biggest improvement margins.  I can spend my time practicing a frame perfect jump that saves me 1/2 a second in my speedrun, or I can use that time to find a path that saves me a full minute over the course of the game.  If I can add a fourth speedrunning tenet, it's that solid overall play trumps highly specialized tricks.  Practicing the toughest tricks that only save a tiny bit of time should be secondary to the ones that save you lots of time.  In VVVVVV, there's a trick that saves a good eight seconds if you do it right, but can cost you twenty seconds if you mess it up--if you do, you might as well start the run over.  When choosing between learning that trick, and one that's stupidly hard and saves me only one second, guess which one I pick.  If I had to draw a Melee comparison, it would be the difference between learning to wavedash consistently and learning how to shine b-air people with Fox.

As time goes on though, if you want to continue to have the best record, you have to make your record *better*, because other people will try to improve on it.  So you'll need to optimize and improve in places you didn't know were actually possible, and constantly reinvent your own speedrun.  You may need to take tactics from other people, which is totally legitimate.  It's not about inventing tricks, it's about having the fastest run.  Developing a new method for playing the game can *help* you get the best time, but if you adopt somebody's technique and do it better, more power to you.  I don't care if my speedrun looks just like everybody else's, provided it's the fastest.


Bonus section!  So, I also promised on the IC boards to write something special just for IC players.  Here it is!  Some of it's going to be a little obvious, some of it less so.

Your Guide to Not Letting Nana Get You Killed

Nana is great.  She's your best friend, the source of half your damage and probably more than half your KOs.  There's a problem though.

Nana's dumb.  Very, very dumb.  And like somebody who is dumb, she does dumb things.  I've become very acquainted with just how dumb she can be.  So any time you're going to try and interact with her, remember: she is stupid and will try to get you killed.  So here's a list of things that you should keep in mind while trying NOT to let her kill you.

1) You cannot save Nana while she is tumbling.  If Nana has lost her double jump and goes into a tumble, she's gone.  Forget about her.  Killing me won't bring her back.

Okay, that was stupid.  Point is, you can't save her from a tumble, so don't try.  One important If she gets shined by Fox however, that breaks her out of her tumble, meaning you can belay and teleport her to you.  The problem is that there's a Fox right near the edge and belaying leaves you open, but then again, going SoPo against Fox is eighty kinds of annoying, so the risk might be worth it.

Nana cannot break herself out of tumbles except with her double jump.  She double jumps the moment her character model becomes level with the main surface of the stage (except on Yoshi's Story, where she will jump towards either the platforms or cloud, depending).  If you practice in training mode against another ICs and just hit them off the level, you will soon learn the exact timing for when she jumps.  The moment she uses her double jump, you can belay or forward+b, so keep that in mind.

2) When Nana is in hit stun, she will not belay or forward+b with you.  Even if she is not sent into a tumble, she won't come to your aid.  If you were to up+b at the exact frame she gets hit by the weakspot of Luigi's up+b, she would not teleport to you.  Don't rely on her to belay and save you if she's about to get hit.  That said, if you can snatch her away from getting hit by using your up+b, it could be worth your while to do so.

3) Nana will not do anything for you while she is teetering on the edge of the stage.  She will not up+b with you, she will not forward+b with you.  I think you can force her to act if you hit the c-stick but that's it.

4) Sometimes it's best to let her die.  If Nana is really far out there and the opponent has a free edgeguard situation, don't bother helping her.  It will get the both of you hit.  I'm sometimes criticized for not helping my Nana enough, but the problem is that most of the time, belay will leave the both of you wide open.  Either she'll die, or you'll eat something really deadly.  And it's super tough to aim her so she'll grab the edge, so half the time she won't even make it on her own.

This is where the up+b ledge-cancel comes in handy.  It lets you save her when she's kind of close to the stage without putting yourself at a significant positional disadvantage.  She becomes invincible above the level, and you get edge invincibility too.  Even if she gets hit by the opponent, you're in a prime position to punish them for it, so at least you get a trade.  Remember, you lose a stock when Popo dies, not Nana.  It feels like the stock is over without her, but you can still attack, deal damage, land stupid d-smashes, and chaingrab a bunch of characters with SoPo.

5) Nana has a special platform AI.  When she is on a platform or above it, she goes into a different AI pattern than normal.  You cannot up+b and collect her if she is in this mode, even if you are close to her.  Try falling through a platform then up+b'ing immediately; she won't join you.  If she's stuck above a platform for whatever reason while you're trying to recover, do NOT try and belay because she won't do anything to help you.  If you want an example of this, stand on a platform, drop down and immediately use your up+b.  Popo will belay, Nana will do nothing.  If you wait for a few frames, THEN you belay, she will go with you.

That's all for today.  Thanks for reading!


  1. I remember seeing trailers for this game (VVVVVV) at TIG and such; looked interesting. Do you do speedruns for other platformers? (i.e. Matt Thorson games, Cave Story, etc.)

  2. Seee

    Wobbles is like

    info man

  3. What do you think about using emulators to do speedruns? Does that take away from the thrill of the speedrun in your opinion?

  4. 1st: Nope. Probably won't either, VVVVVV just kinda spoke to me so I wanted to. I haven't touched it in awhile though, got kind of hooked on League of Legends.

    2nd: Yep.

    3rd: Not especially. The emulator itself lets you use save states which is really handy for practicing. If you mean TAS, then that is interesting in itself because it shows what is technically possible. It's not thrilling, so to speak, but it's interesting.