Behind the Reports: Musings on 500 MS Bolds

[Ihavenoname248]

I'm still letting that sink in. 500 is such a small, unassuming number as a number. It's a fairly short period of time in seconds, being just over 8 minutes. You can't do very much in 8 minutes.

You can do a lot in 8 minutes (complete all but 5 or 6 official CC levels). It's an incredibly long time to have to make 0 mistakes during (Warehouse II, Cityblock). And when taken one item at a time, 500 is a veritable mountain. I'm not done optimizing, nor am I done with any of these scores (ok maaaaaybe done with CCLP1), but this is a good point to take a hiatus. These past dozen bolds have all taken a lot out of me and I definitely need the break to work on other pursuits, so for now I will just reflect on what it took to get here. (as well as the Chip Cup from the past 2 years, since I need to average 7th to lose this year... heheh) 500. Wow.

February of last year, I scored 700 on Cake Walk to take 10th place in the original set and claim the Bit Buster award. I had 403 bolds at the time. Trying to go chronologically from here would be confusing, since I don't remember a lot of the specifics, but I do remember Ruben gained 1 second each on Teleblock, Lemmings and Jailer to reach 5,977,030 points and knock me out of the top 10 (November 30th). Naturally I refused to let this stand and went to improve my 413 Blobnet. That same day I spent a couple hours playing the level, at first copying Miika's 422 safe route directly and later using judgment to skip waits (but still no risks). This resulted in a quick 427, settling me firmly in 7th for the time being. But then I realized I could take a couple risks early on, cut a few corners and before long reached 431, meeting my goal of 430+ for the level. I am still yet to improve Block Buster or Cake Walk, but I'm much more confident in my ability to get a solid score now. The new goal is to pass Andrew Bennett on all sets: this seems doable with significant effort, but it feels like it can't be a bigger mountain than 500 bolds from 0 in this span of time.

CCLP2 I'd gone through and picked off the easy bolds while playing through for the first time. This left me with some fairly significant execution challenges to overcome, and I made the mistake of sorting these into the same classifications I gave CC1 levels: Definite, Probable, Unlikely and HAHAHANO. Funny aside about HAHAHANO- originally CC1 had 4 levels in this category, being Blobnet, Blobdance, Cityblock and Doublemaze (I could see Block N Roll being something that happens at some point as 433 takes very little reacting). Now there are three, as I not only found 549 on Cityblock, but found and scored 550, beating pieguy!!! I'm not convinced my route is optimal, but I'm in the CC1 MS evolution page now anyway and nothing can take that from me (of the 500 bolds I have this is one of the ones I'm most happy with, and if you want to match it...well, don't be afraid to break the patterns.)

Anyway back to CCLP2. The Definite pile was fairly small and easily scored. The Probable pile had a couple naive placements, like Jungle (seriously me?) and Island Hopping (only took about a half hour or so, wasn't too bad). Working through these didn't take too long to reach around 120 bolds, at which point I hit a wall. Oorto Geld II was a pain and I had to score it twice as the first time I missed the -.9 and got a -.0 instead. Iron Mysticus and Key Color also both proved much more challenging than I expected (Key Color in particular with a 1/10000 string of bad luck with the blob cloner blocking me the first time 10 times in a row...), but as usual persistence in attempting and practicing specific sections helped significantly. Wrongly pegging difficulties of levels continued as Icy Moat fell quickly (I expected to struggle with the timing, but it was close to 49 Cell which I already knew) but Madd Maze was a full day grind, with an abyssal RFF room success rate (my route through has an extra input compared to the AVI or TWS routes, but this extra input helped me as it meant I had a consistent input rhythm), and then the blobs kept directing the gliders wrong >_<.

Eventually I worked through all the easy levels, and so progress slowed significantly at 480ish total (I'd been working on other sets during the CCLP2 focus, but not as much). Pauseless Captured and first try traps Gauntlet helped, as well as saving a bunch of long sokoban routes for all in a row. After Cityblock, I moved to CCLP3, scoring two new records (Countdown with Hornlitz and Get a Clue solo!) and copying a bunch more routes. Checkers and Flame Thrower were two non-public routes I was able to find with minimal difficulty as well. Flame Thrower in particular was a lot of fun to work out a route for in the editor, and then I just adjusted when I would pick up certain chips and scored 370.6 immediately. A small tweak to the last room to trade [1] for [2] and a chip was all it took for the bold of 371. I also found the Alphabet Soup route a few days before scoring it (thanks James for pointing out that my 948 was the 949, but missing a move in execution which snowballed >_<).

Three days ago I got access to MSCC. It was the second most frustrating experience I've had with this game (I'm getting to the most frustrating, don't worry) as Every. Single. Input. was delayed by an entire move. I had to count out how many times to press the key individually, then not screw up the rhythm and on the longer walks I was counting "1-2-3 1-2-3 1-2-3 1-2-3 1-2-3" or "1-2-3-4 1-2-3" or something else like that. The actual teleport skip in Graduation was trivial. This was bold 493.

The previous two days were spent on long sokobans. On the Rocks and Mind Block were both very simple to score, especially after noticing Mind Block had a consistent pattern and 'merge point' in cloning another block. One of the biggest route memorization aids I've found is finding a place that must be visited between phases and remembering what happens between those visits to that point. For Warehouse I, the next bold I scored, that was dropping a block into a bomb. Executing and remembering the route took a couple attempts, but was ultimately much easier than I anticipated. Yet Another Puzzle took a few more tries than I expected but was still fairly easy.

And then we got to Warehouse II.

I tried scoring Warehouse II back when I first played through CCLP2 last year. It didn't end well and I took b-191 initial solve. I expected a repeat, but all the maneuvers made sense to me this time around (thank you, optimizing Shifting Maze out to 832!) and I was able to score the perfect decimal after only about 60 minutes! I expected it to take much longer than that. Oracle II on the other hand, was an interesting blend of everything thus far. Long opening (2.5 minute first room, very easy to remember) and then a more complex second half that I kept making small mistakes during. I was still able to score 598 after another hour or so, though it nearly gave me a heart attack when I lost 3 moves on the final trip up from the bottom! (oof, oof, missed spring slide)

And so I reached 499 MS bolds.

All the way back at 470ish I decided that number 500 should have an appropriate title and you already know it's Which One Next?

This was a dumb decision. I regret everything. This was the most frustrating experience- not for remembering the route, no, that was easy with a carefully notated map ( http://i.imgur.com/VldpXwA.png). All of my frustration here came from execution, as there is really nothing hard in this level (other than the 11 teleport run about halfway through). But I kept screwing up the simplest things, and it took me 3 hours of attempts (over the span of a lot longer than that) to realize I was trying to teleport too fast. Slowed down the pace, found the right rate, and had a flawless run going all the way until 30 seconds left, where entered a station from the wrong side. Then I had another perfect run to the very last teleport sequence, where I still went too fast. Finally, I managed to not screw up the ending after losing [1] in the 11 teleports to take my 795 and 500th bold.

I am not looking forward to Oversea Delivery after that.

Anyway, thoughts on a couple levels and stuff!

CC1 levels

Lesson 7: Taught me how to boost sort of okay. I kind of brute forced attempts until scoring bold 20+ months ago.

Trinity: Taught me how spring slides sort of worked. I kind of held the key every time and hoped to change directions right. Now I know to double tap things.

Hunt: Taught me that following every step exactly the same as a reference route is a recipe for dumb mistakes. I copied the beginning and then just made up a path for the rest as it really didn't matter.

Blobnet: Much much later, taught me how to spot whether blobs were on their moving turns or their non-moving turns.

Blink: Taught me how to spring slide properly- with the double tap and then direction change. I do still hold sometimes for when all the direction changes are equally offset, however.

Arcticflow: Taught me that practicing specific areas was much easier with a copy of the level edited to only be that area. It still wasn't an easy bold at all.

Mishmesh: Taught me how to play off a map. This has come in handy a LOT.

Seeing Stars: Taught me that pausing to help remember routes was viable. I've since honed this quite a bit farther, but for the sake of real time I rarely use this technique.

I Slide: Taught me the value of good notes. This has also come in handy a LOT, and not just with simple moves. Combining an encoding, textual instructions and a map on the side is how I execute routes with a lot of variety and backtracking.

Spooks: Taught me that pieguy is unsurpassable, and sometimes unreachable. Also taught me that J.B. is nearly on the same level.

Four Plex: Taught me that the publicly available route, even when it's the bold, isn't always perfect.

Cityblock: Taught me that pieguy is beatable, hooray!

Mixed Nuts: Chips under blocks are apparently my specialty and I learned more about how spring slides work to save time here. Still not enough for a new record.. :/

Mix Up: Taught me I could remember long sokoban routes. It was around here that the reasons why certain orders saved time started clicking for me.

Yorkhouse, and to a lesser extent Catacombs: Taught me not to trust Andrew Bennett's probability calculations. They always err on the side of optimistic, as Catacombs is 1/655 and not 1/455 and Yorkhouse is way below 10% odds. The grind was annoying, but...

CCLP1 levels

Graduation: Taught me that MSCC is garbage and how did I ever play with this and how did anyone ever put up with this and how does James even manage now it's impossible. Also, Teleport Skip Glitch is easy someone add to Tile World please so I can score Skelzie.

Wedges: I can blocks!

Tetragons: RFFs aren't that bad...

Square Dancing: Always test your routes before counting moves, and always count the free first move when counting moves.

Descending Ceiling: If something seems weird about the scores people have, there's probably a couple tricks to find and each person missed a different thing. Applied to Get a Clue. (hint, hint)

Who's the Boss?: Shift-O is dumb.

H2O Below 273 K: holy crap this game gets hard to execute fast wow (273 -.9 still very happy with this thanks)

Mini Pyramid: Yeah, chips under blocks are definitely my specialty.

Chip Kart 64: Note to self: sliding less tiles is faster than sliding more tiles due to more overrides and boosts.

Colors for Extreme: Keep an eye out for blocksliding opportunities. Technically applies to Booster Shots as well, but this came first chronologically so

Bowling Alleys: Just because a lot of the top players have the score doesn't mean it's optimal. Also applies to Get a Clue, but this came first. Also, just because you barely scraped another second out of the level doesn't mean that your decimal is optimal.

The Shifting Maze: Yep. I can definitely blocks. Shuffling everything around, more loops, small timesaves, small "hey I don't need that!" moments... this is probably one of the most complex CCLP1 routes to develop.

Time Suspension: When grinding RNG, set aside an entire day, have other things prepared to keep from getting bored, and the pull of a new record can outweigh a lot of tedium/boredom. Also, walkers are worst monster. By a lot.

Portcullis: You can slip by the single ice tile clone thingy in Lynx. Huh.

Easier Than It Looks: Ok seriously if the same route hasn't scored a second higher yet it's not going to without a change. Also, pay attention to teleports and where they lead. This can shortcut (hi Countdown!)

Cineworld: Half waits suck.

CCLP2 levels

Use the Fish: Taught me to just boost for it when failure means death: going to slow means a restart in this setting!

Madd Maze: RFFs suck. Just thought you ought to know.

Just Enough: Tedious, tedious, tedious, so glad I get to copy a route instead of have to iterate the timing of everything myself...ugh.

Roller Coaster: Just because the AVI is perfect doesn't mean I have to be perfect. Think through if moves can be lost, and if so, where do they start mattering. This goes double for the TWS losing 0 to the blobs but 6 to the walkers. I lost 8 to the blobs and 0 to the walkers in my run, since the walkers will often either let you past or kill the attempt.

Monster Factory: Sideswipes exist. Better routes for travelling salesman problems don't.

CircleMaze: Pay VERY close attention to heavy boosting sequences. 475 takes very little luck and 476 does not require >R (override) >R (override) >R at the end.

Captured: Mouse clicks aren't so bad...

Gauntlet: Random results will behave differently for everyone. I knew this already, but James mentioning how bad it feels getting stuck in the trap on 458 potential had me expecting it to happen on my first try past the walkers, so I was pleasantly shocked when it didn't!

CCLP3 levels

Lot in Life: Illogical routes sometimes trump logical routes. My poor 95 MS....my 95 Lynx though is pretty logical. Not easy to find without the right insight, though.

Map the Path: It pays to be in the right place at the right time for the scoreboards. I was around when J.B. found 245, and was able to find it before anyone else came online. Looking back, I think this is the moment I realized "yeah. I can actually do this optimization thing."

Recess: Even bad levels make CCLPs sometimes.

Spiral: See previous level comment.

Blazes: James gets ridiculous insights with monster interactions. Pay attention to them.

Get a Clue: Pieguy is not perfect. I am able to find things he missed sometimes. (This, combined with the imminent J.B. look got me to take a look at Cityblock and score 550).

Flame Thrower: Just because one person found routing the level easy/hard does not mean I will find it easy/hard. J.B struggled, I had no trouble. Conversely, James found 825 on ACD and I still don't see how an improvement isn't 826. I suppose finding a "half cycle" on the balls would be the key.

Once Upon a Troubadour: Moving mouse clicks are a pain when boosting is involved. At least it's short.

Countdown: Just because bold has been reached doesn't mean you should stop looking. Shoutouts to Shane for finding the trick I missed to raise 71 to 72, though if we hadn't, triple joint bold with J.B. would have ensued

Checkers: Miika is really good with blocks. Listen to his ideas. In Lynx, slaps are awesome, splash delay is not, avoiding splash delay is.

Mediterranean: Collabs are awesome. Long teleport runs are not.

Which One Next?: Ok, what did I just say? Oh right, long teleport runs are not awesome. This is a bad level. Yay 500 bolds with appropriate title (and 501 has a recommendation already).

Lead Us Not into Temptation: Walkers are still worst monster. Also this is more likely than Time Suspension, which is still unconfirmed. Despite the route having been public since I scored it.

Manic Depression: Sometimes, just sometimes, randomness will go your way the first chance it gets. Also cross checking into rams is awful and the first bunch of this level is a huge pain.

Mr. McCallahan Presents: Some busts are convoluted and use every part of a level and several obscure mechanics. This helped me fix Fish in a Barrel before anyone else caught the ridiculous MS only bust in UC4.

Chip Cup

I would have had a real shot in the 2014 Cup had I been around to enter the super competition...oh well. I still did well for what I was able to enter. I went into the 2015 season aiming to win, and so far I've built up quite the lead. Got sniped on the pieguy award with the survival competition last year by pieguy himself. Still haven't had a single treasure hunt competition to enter. Sometimes, the rule that better Lynx scores are counted as equivalent is silly when Lynx is clearly faster by a significant margin. The Power of Slide Delay Compels You! is a great level and I had a lot of fun trying to make everything I wanted to do with it fit into the level, as well as tweaking the slide delay rooms. J.B. is not infallible with optimization and can be beaten in time trials (I did so twice!). If a level gets updated, always look over the new possibilities.

The past 22 months have been a ton of fun. Here's to another long period of time, more intricate routes, more fun levels and stellar community packs! <span style="font-size:8px">_{I'm coming for that CCZone Awards Leaderboard overall first}