At first glance, Pictionary for NES appears to be a board game, not acceptable for submission to this site. However, per Masterjun's rejection comment in Spikestuff's submission: "It also has video game elements... However, these are skippable..."
I had the same thought about this game. So I said, "What if we didn't skip them?" and created the TAS contained in this submission.

Emulator used: FCEUX 2.2.2

This improves upon my previous movie by over 6 seconds.

In order to justify the existence of this run, I will be contrasting it to the alternative mode run, which, I think we can all agree, sucks.

In alternative mode, you are expected to pick a card from your real Pictionary deck, then draw the word on your NES, and get your friends to guess it. This drawing section can be completely skipped, after which it asks you who won...to which you can always reply yourself.
In short, you keep telling it you won until it tells you you're the best. Even if you don't skip the drawing portion, what should you draw? Who would guess it? This mode has very little potential for entertainment or improvement.

Regular mode is a little different. Rather than draw words for your friends to guess, you are given minigames that help you guess various pre-drawn words (at least in singleplayer mode). Even if we skip these minigames, regular mode is still a more interesting run than alternative, because you have to type in the words. The input system is surprisingly annoying nuanced; you end up having to hit all the letters in a 2-frame window along the way, which looks impressive to anyone who's stumbled through this awkward input system themselves.
Another interesting aspect of this run is that the words are generated in a pre-determined sequence, which is chosen when you press start to enter your team name. This means that RNG is a little more complicated than always manipulating the fastest word ("NO"). DrD2K9 and I worked together to figure out which sequences are faster; I ended up writing a lua script to aid in this process. However, it doesn't actually simulate any inputs! I chose to trust its results, but it's possible that, through manual testing, a faster sequence will be found.

As an interesting aside, the sequence that realtime runners grind for ("NO", "PIN", "THINK"...), is 285 frames slower than the sequence I chose for this run ("TWINS", "SWAN", "COMEDY").

Also bear in mind that there is a 128 frame rule after entering each word. Any potential improvements to this run will have to deal with those.

For this run, I attempted to execute the fastest death in each of the minigames. Here's a brief overview of each one, in order of appearance:

The first and last spaces (red ones) are random, so we can manipulate any game we choose. The paint one is by far the fastest, so I choose it both times; that is why you will see it 3 times throughout the run. There are 6 spaces to land on, but only 4 minigames, so some repetition is inevitable.

Watching this run, you may believe this particular minigame is about a paint bucket collecting green paint drops. This is not the case: those green drops are actually bullets. This is a space invaders knockoff, and the bucket is suicidal. RNG determines when and where the green drops get shot, so a faster death may exist here.

Between the first and second appearances of this game, I show off a weird glitch: if the paint can is tipped over when the game unloads, it will still be tipped over when the game loads the second time. It's visible for only 7 frames starting at frame 2315--very hard to notice if you're not looking, but do try to spot it if you watch the run!

This one is my favorite! It features a disgruntled employee who sets out to destroy as much company property as possble. Deciding when to walk towards the little gremlins, and when to turn around and grab more boxes is an interesting challenge.

I HATE this game! It's a perfect example of extremely poor game design. These clones you have to save appear purely at random; I believe there is no configuration in which you can save all of their lives, although sometimes it's just barely possible to fully uncover the image. This game basically gives you a goal with no way to achieve it, which is the absolute last thing a game should be. This minigame is part of what motivated my decision to do a fastest death TAS over a full completion TAS. Hopefully you'll enjoy watching these firefighters fake out the citizens and dance to the music instead of doing their job.

This one's fairly straightforward. You just have to walk back and forth, letting the pink guys scuff your little spaceman boots. I'm almost certain RNG doesn't affect their behavior any, but perhaps I will be proven wrong.

I touched on this earlier, but here are some more details. The word sequence you get is determined as soon as you press start after choosing a team name. This means that any run--one that improves upon my minigames, one that does them way slower, or even one that skips them altogether--will have the same sequence of words. This effectively makes it an entirely separate problem to solve, independent of any specific run.

The text entry frame rule interacts interestingly with dice RNG. It turns out that you can wait several frames to type the final letter, still make the same frame rule, but somehow end up with different RNG. This makes a first frame 6 roll always possible! Here's my theory on why this works: Pictionary probably has an RNG variable that updates every frame, except lag frames. Whatever arcane RNG math gets performed takes into account the current frame, as well as the previous value. After sitting through different numbers of lag frames, even though we make a roll on the same frame, the formula is looking at a different previous value in each situation, and so the results are completely different than before.

If I seem to waste time or go slower than I could, this is why.

If you've ever played this game, you'll know that the text input system is very...unique. Rather than give you a standard alphabetic grid, or an arcade-style letter select (like at the beginning), they give you a pair of high tops to stumble over the alphabet. These shoes actually run. They are subject to acceleration and friction. At full speed, they travel at 1 letter per 4 frames, giving you a 2-frame window to type each one--very hard to pull off in real time, but pretty simple to do in a TAS. To keep things interesting, I type out different words and phrases, which I backspace along the way. My personal favorite is after the second paint game, where I'm able to type 4 out of the 5 English words that start with SWA-!

Sometimes, it becomes necessary to turn around when entering a word. You'd think you could stop on the letter you need, then start accelerating in the other direction, but you can't. You can only stop on even letters (B, D, F...), so sometimes you have to overshoot. Usually you can overshoot to the next even letter, but sometimes you end up overshooting even further. In either case, it's always faster to overshoot and then accelerate backwards, than to stop early and inch forwards. This has the effect of making the input look a little bit sloppy, but I assure you that all my text input is optimal (aside from the RNG framerule thing).

6600: The game looks at a left arrow and comments, "Nice going, that's right!" I used this for my YouTube thumbnail.

Pretty wordy for a 2 minute movie, huh? I probably could have spared you some of these details, but I don't feel like they hurt.

Anyway, I had a lot of fun creating this TAS! Probably the most fun anyone will ever have with an LJN game...
In some ways, it turned out to be much deeper than I expected. And of course, Tim Follin's incredible soundtrack provides an exciting backdrop to it all.

I'm well aware that this is an iffy game for TASVideos, so I'm curious to see what the community's response will be.

Noxxa: Judging.

Noxxa: This is not a Vaultable category, so this movie would have to pass for Moons qualifications. For that, I'm not really seeing sufficient support accumulated from votes and comments, with the "gameplay", such as it is, not showing a lot of merit - particularly because the minigames are not actually played, just ended as quickly as possible (short of outright skipping them). In the end, there's just too little appeal for it to really be worth a Moon. As such, rejecting.