Barrier Chains

[@Enderdude600]

Very cool but I think theres one other glaring issue I found: Because of how the line snakes up and down the screen, looking at the middle does not make it immediately obvious if an entry going up/down is an increase/decrease in position.

[@cubingguy9628]

Maybe you could use a hilbert curve for this

[@MrEel-dc4kh]

I think replacing barriers with arrows or triangles might fix this!

[@0-Kirby-0]

​@@cubingguy9628Literally my first thought once I understood the premise was "ooh Hilbert Curve!" though honestly I don't think that'd be much clearer. Sure you can tell something is making progress if it's vaguely moving counter-clockwise, but you still lack any real instinct within quadrants. It's just really difficult to represent a linear relationship in 3d without wasting a dimension. My best idea would be to separate the data set into categories, like orders of magnitude. Higher rows are a higher category, each sorted left to right. That way if a cell is above another, there's a clear instinct of meaning (it's at least 10x higher), and if you want to follow the overall hierarchy you read left to right, top to bottom, just like most text in the West. This does mean you need more space, as categories won't alway contain the same number of elements, and you have to be able to pick sensible categories.

[@ujocdod]

@@MrEel-dc4kh I agree!

[@tbouchard2789]

Instead of "Stagger-snaking", Its prob EZier on the eyes... to shape your chain, into a rectangle, or circle.... Riding edge of screen.... Spiraling toward Center-screen (which would be the, Top-of-the-list position )

[@WTIF2024]

Cary kitchen hacker is back

[@hooplow]

Dare I say, Cary KCary HCary as well

[@maxdoesstuff2179]

@@hooplow cary kidnaps human

[@YoylecakeVibes]

Let’s gooo!

[@dogg0nit32]

Cornelius Archibald Rutherford Yates Kensington Huang my beloved

[@kgratia4748]

MY KITCHEN IS HACKED

[@Aerotactics]

The barrier size is hard to read, I think changing the color of the barrier will help with that visualization: Red = big gap, Green = small gap, yellow = mid gap.

[@nonagonguy6121]

I agree. Changing the size of one of the sides instead of the total area would make it easier to read aswell.

[@Flourish38]

This! Also, discretization often increases legibility. Something like

[@weirdlyspecific302]

The entire point of having the bar is to be able to visualise how much two adjacent entries differ size in non-discrete manner. To make it discrete would negate the whole point of having a visual representation that highlights >>subtle

[@Flourish38]

@@weirdlyspecific302 If you think you can reliably tell the difference between a 4% gap and a 6% gap with this current method, then be my guest. If that was the true goal, it should at least be logarithmic.

[@wack1305]

All it needs is the bar to be one dimensional. Visualize comparing areas and intuitively converting them to percents is harder than something like a progress bar we’ve been trained to read

[@dawica]

As a data viz nerd, here’s my criticism: 1. What you’ve effectively made is a relative bar graph (i.e. a bar’s size is equal to the data point’s difference to its neighbor, not its absolute magnitude) with a huge label, made the x axis super skinny, made the bar grow in area rather than length, then columnated it. Because it’s essentially a modified bar graph, it actually has the same scaling problem as bar graphs, just with more room before it gets crowded. 2. Having the graph snake adds a severe readability issue in that you can’t immediately tell whether a column is increasing in the up or down direction 3. You say that the bar graph wastes space, but it doesn’t. That empty space in the lower right or block of bars in the top left is what gives the graph its the visual ratio. You aren’t “wasting” the top 3/4 of your fuel gauge when your car is on empty. That might sound pedantic, but almost all of my criticism is due to the visual information that is lost by “removing the wasted space”. For example, because the barriers only focus on the delta, it isn’t visually obvious whether the gap is closing because the smaller value is growing or the larger value is shrinking. 4. It’s hard to gauge a value’s rate of change 5. If the purpose is to highlight when two entries cross each other, the barriers should grow as the gap closes (maybe make the size scale down by a factor of smaller / bigger, possibly logarithmically to emphasize close races). I know that breaks the “barrier” metaphor, but you want data to become *more* visually apparent as it becomes more interesting/important. 6. It doesn’t do a good job of showing an absolute run away success. The difference between being 10x your neighbor and 100x your neighbor is only a 9% difference in barrier size

[@ADudeWhoDo]

For 5 you could change the analogy from barrier to path/tunnel and it would make more sense because making it bigger gives more room to cross over

[@dinhero21]

1. It shouldn't have the same scaling problem as it belongs in another asymptotic complexity (O(n) for bar graph vs O(n²) for barrier chain)

[@BryanLu0]

I think it isn't visually obvious whether the gap is closing because the smaller value is growing or the large value is shrinking in a bar graph race either, especially when the two values that are converging are isolated from the rest of the values

[@deamichaelis1]

You could toss a mirrored bar graph (sort of like a violin graph) along the snakelike path to show the differences better. Thus it is more of a visual identifier of the difference and what way is up in the chart even when looking at the middle because it would go towards the bigger end of the snake. This would be behind the barriers and the images and laid ontop of the grey path, and if you think that would make it to busy just make it white on that gray background thus it isn't to overpowering.

[@ADudeWhoDo]

@@BryanLu0 maybe the barrier could be changed so that the sides are angled towards the value which is changing less, kinda turning it into an arrow that shows whether the change is from values going up or down

[@DataVisuals99]

Very interesting graph! Although I do slightly prefer bar chart races, I like how aesthetically pleasing it looks and it has so many cool elements to fix up the problems with the bar chart.

[@puzzLEGO]

its definitely an improvement on normal chains, but as you said, its not entirely intuitive of the absolute size, so you can really only compare directly adjacent elements. perhaps try playing with size or colour to show each element's absolute value (apart from the number itself of course)

[@color_rainbow_2]

NO WAY VERIFIED KZbinR ON CARYKH

[@Crazyclay78YT]

i know, i was thinking you could have like a "scale" down in the corner or something where it says like red = 10 difference, orange = 100 difference, etc. or something like that

[@gownerjones]

I find it hard to even compare adjacent items here because there's no intuition between the size of the barrier and the ratio between the items. I think the barriers would be better if they were similar to the old youtube like/dislike indicator where there is a fixed length bar split into two colors. One color is one item and the other color is the other item. As the value of one item becomes relatively higher, its "half" of the bar grows while the other shrinks. That way we would see in an instant what the relationship between neighboring items is and we could still anticipate the moment they swap.

[@MaxricoQuen]

Somebody im subed in: Cary K rill H elper.

[@patrickskelly8517]

What if instead of a barrier you connected each square to the next with a trapezoid? Where the top of the trapezoid is the full width of the top square, and the bottom width : top width ratio is the ratio of the values. So if New York had half the population of California, they would be connected by a trapezoid half as big on the New York side as on the California side. In the limit of two values being the same, it becomes a rectangle, and in the other limit it becomes a triangle. This would work for your clustering example, you'd see squares in a row with almost equal width between them, then a "neck" in the snake before the next chain of equal widths.

[@swellowtail3117]

i really like this idea!

[@faland0069]

i thought about this exact idea, then i scrolled a little more down to see someone had already thought of it before me lol this would also fix the problem where if you look in the middle, it's not immediately obvious which way is up and down on the leaderboard

[@mathdj2654]

I had similar idea but instead of trapezoid it was inequality sign '>'. The bigger the ratio, the wider its legs stretches away. Though trapezoid one would look more neatly!

[@Gamesaucer]

Ooh, this is a very nice idea! This makes sure it's immediately clear what the ordering is because it looks like an arrow!

[@mk__cyanheron1154]

Rather than a barrier, a trapezoid would be more obvious choice to show relative size. Then the rectangle would mean equality, triangle absolute domination. Plus ratio could be transitive. Having one side of the trapezoid get bigger and bigger when moving throigh the snake. In order to avoid infinitesimal sizes this could be transitive at the span of the column (first in column gets full bar, last gets one pixel bar) and it is explained just like in those videos about scale (bag of 1$ is 1k$, lorry of bags is 1M$, carpark of lorries is 1B$ etc,)

[@mk__cyanheron1154]

Plus direction of the trapezoid instantenously shows which one is bigger (in which way the snake increases in this column ) when looking at the middle of the snake

[@battlepig1014]

discord pings at 2:51 really confused me 😭

[@Crazyclay78YT]

i know right, i had to rewind to make sure 😂😂

[@Thomasthetackmasterpersob]

I like potatoes

[@40Ccents]

_bd_ _bd_

[@razkrispies]

wow cary. you are the best Ned Flanders drawer of anyone who has ever drawn Ned Flanders. I will hang this up on my wall. thank you cary. I am currently sobbing.

[@GameyGaming]

I wholeheartedly agree

[@willharris-braun4422]

I think along with a small barrier between the two items, the space between the two items should shrink as they get closer to the same value. Another way to say it is the distance from the top/bottom of each barrier and the adjacent item should remain constant so two adjacent items move towards each other subtly as they get closer in value.

[@phonetyx]

it would be fun to see this with different space filling curves, make it squiggly and funky

[@PaulFisher]

[banging on pots and pans] Hilbert curve! Hilbert curve!

[@nanamacapagal8342]

good luck reading that

[@Sanjay.2133]

CARY CARRYING US

[@art1637]

Cary karrying hus

[@WTIF2024]

SANJAY WHAT ARE YOU DOING HERE

[@amazingkeller]

@@WTIF2024 who knows, unfortunately pinapples.

[@DiggyPT]

cary carrying barrier chain

[@Istamtae]

everyone is here

[@themistakeisintentional-dn5df]

This channel is like the perfect thing to put on in the background while I'm busy with other things.

[@MarylandballProductions]

0:10 Maryland mentioned 🗣️

[@hirandompeopled4968]

MARYLAND WOOO OLD BAY BEST BAY

[@olbluelips]

You can always tell a place is irrelevant when the people who live there freak out every time it’s mentioned

[@derrikwalker7548]

@@olbluelipsyoure irrelevant

[@SaraMekonnen-c8q]

iMovie HD 6​@@olbluelips

[@what_27]

@@olbluelips yeah, that’s why I freak out when NM is mentioned- no one gives a crap about it, even compared to maryland

[@hi_imlolo]

ive been rewatching so many of the carykh videos recently and now theres a new one after so long thank u cary its like u knew i just wanted more

[@zhuzhou]

I love it Cary! Feels like youre onto something. I expect there will be a ton of work innovating around exactly how to draw barriers, etc here and this really could take off!

[@THTIMT]

Cary, you are such an amazing data analyst! Congratulation for inventing and sharing the new type of data visualization! Your creative data representation will truly inspire many statisticians and BFDI fans!

[@Pulsars]

carykh uploaded, now my life has meaning again

[@NickCombs]

That's a cool experiment! I'm inspired to take another approach to the same goals. The main idea is that the highest value is displayed as the largest item in the middle, and every subsequent value will be a proportionally smaller that spirals outward. This should solve two issues: 1. The biggest value is somewhat hidden in the top left corner and doesn't stand out above the crowd. With it taking center stage, it gets the attention it deserves. 2. The barriers don't convey the ratios, like you mentioned. When we scale the items instead, we can see that info more intuitively. I do see a possible issue with small values being too hard to read. However, I don't know if that's necessarily a big deal if we're only interested in who's winning the race.

[@NickCombs]

Oh also instead of forcing items into a spiral which might not fit as well on a rectangular screen space, we could let them float around with attraction and repelling forces like we see on node graphs.

[@Hanniah--Heart]

Dude i've been trying to find your channel again for years! I've missed you! You're possibly my favorite mathematician

[@Vioviovio_341]

THANKS FOR MAKING MY DAY CARY!

[@Twin138956Productions2019]

CARYKH IS RETURNING WITH THIS ONE 🗣️🗣️🗣️🔥🔥🔥

[@alexanderdinkov8002]

Here's the problem I see (and how to fix it): - The snake-like twisting is quite uncomfortable. The reason is that If I look at a random square at the middle of the screen, I cannot instantly tell if it is bigger than the square above it or smaller. Sure, there is a "barrier", which tells me what the ratio between the two is, but this ratio is useless if I don't know which of the 2 is the bigger one. - Instead, in order to figure out which of the 2 is the bigger one, I need to trace the snake with my eyes until I see the nearest turn. Then I need to figure out what that turn means in relation to the direction of the square. Then I have to move my eyes back to the squares I was looking at initially (but I might not find them instantly, or they might have swapped while I was looking away). - A chart where you have to look away from a piece of data in order to understand that piece is not a great chart. So lets fix it: - The easiest possible fix is to simply turn the background black. That tiny change will make it a lot easier for the eyes to figure out the direction of the snake. If there is contrast, the viewer will need less time while looking away from the square. - A slightly more involved change is to add arrows across the body of the snake. Make it possible to figure out the direction of the snake without looking away from the square that you are tracking. - But there is an even better way. Replace the "barriers" with something that's kinda like arrows. Replace them with trapezoids. - Lets imagine that you have 4 squares: A, B, C and D, with values of 1, 4, 40, and 41. - The space between A and B will look kinda like the text "A>B", where the lines are drawn from the 2 corners of A and form a trapezoid shape at the center of the nearest wall of B. The trapezoid will look kinda like an arrow. The A side will have a length of 100%, while the B side will have a length of 25% (1/4-th). - This is visualization, not mathematics. The ">" in "A>B" doesn't mean "is bigger than". Instead it means "is behind", because ">" is just an arrow-like trapezoid. - For "B>C", the ratio of the trapezoid shape will be even sharper: 100%>10% - For "C>D", the ratio of the trapezoid will be almost 1-to-1. Rounded to 100%>97%. The trapezoid will be almost a square. (So it's a bit more accurate to show it as "C=D" than as "C>D") So If I had to visualize this chart with text, it will kinda look like : "A>B>C=D" The ">" signs (the trapezoids) will make it obvious in what direction the snake is moving. Like... D=C

[@alexanderdinkov8002]

Many people had similar ideas in the other comments. I like to think how that just means that I was on the right path with my own explanation.

[@Erased_UnityXDX]

Oh hey, a new video from Cary Knows Him!

[@hatzyy2763]

knows...who?

[@Erased_UnityXDX]

@@hatzyy2763 him

[@dev0panda2227]

This is such a cool way to visualize data! To solve the issues it has, I feel like entries could be displayed in a spiral rather than a snaking pattern (so it's obvious which direction is increasing/decreasing), and to visualize the ratios between each entry, there could just be a color gradient and/or a smaller bar under them.

[@charlieeeeeX3]

cary kitten hugger posted finally!!

[@expansionpack4485]

Gonna be real with you, when I saw the name of the video, and the outlines of the US states, I thought this was a video ranking the lengths of those states' barrier island chains.

[@bockhssquad6973]

this kind of weird exploration of data communication is the exact thing I'm subscribed for

[@whynotanyting]

Couple of ideas: 1. Use a "loading bar" style within the boxes with the percentage normalized to first place. So if the first place has 19,000 then it's 100% filled, then if second place was 18,500 ≈ 95.4% filled, third place 17,600 ≈ 92.6% filled, 13,000 ≈ 68.4%, and so on. Semantle (web game) uses it and I've found it as a great visual indicator. 2. Color them. Full red for first place then full blue (red/blue for color blindness, you can use whatever) for last on the visible graph, and a gradient in between. Basically the conditional formatting color scales on Excel. You could combine them together with your current set up for the ultimate (albeit overly redundant) graph!

[@MagnettyGirl]

It's always interesting to see new concepts for things that seem like they can't be improved on any further. Who knows, we might get something even better one day. Good job by the way!

[@antonyshadowbanned]

I would totally replace the bar with an obtuse isosceles triangle with the large side toward the higher number and the obtuse corner pointing to the smaller. This way the scaling and directionality might look more intuitive. Awesome video, awesome concept!

[@Micha-Hil]

Casually drops an intuitive bar graph out of nowhere. GG Cary.

[@DevGreg]

Amazing - I can honestly see this becoming a massive sensation!

[@Gamesaucer]

What if you were to use the position of an item not just as a way to indicate its ranking in the list, but also as a way to communicate how close they are to each other? Imagine the following: - You tally up all the "wiggle room" between items in the chain. - You do the same calculation as before regarding how close two items are to overtaking one another, but now... - You distribute that empty space by treating the above numbers as weights. So now if you have a list of barrier sizes of e.g. 0.2, 0.8, 0.5, and a total wiggle room of 10 units, you'd get distances of approximately 1.3, 5.3, 3.3. - You then position the elements so they have that much space between them! This causes two elements to get visually closer to one another as they approach each other in value, which is more intuitive than the "barrier". And the normalising effect of the calculation you do to check how close two numbers are makes sure that there's never TOO much empty space, so that all the empty space that IS there still intuitively communicates something.

[@Puppycakenum]

I’ve always found data visualization and statistics so interesting! Amazing video Cary

[@dillonmccormick8700]

lmao at the discord notification sound at 2:53

[@Chitose_]

yea i thought i got a ping

[@ben9583_]

This is pretty clever. I like how it sets out to achieve these two objectives and uses that sort of as a goal post; it helps convey how this format should be used. Requires a bit of a mindset shift from the regular bar race format, but it's got a nice feeling to it, seeing change take place

[@signisot5264]

My initial response was to make a 3d bar graph, put the tallest items at the back so they're always visible, which lets smaller bars layer over the larger ones and use up some of that space. with a bit of thought, I realized that looking at it top-down it'd look like growing/shrinking squares, which is another easy way to make it intuitive and clear

[@VictorStauber]

You finally back with this channel!

[@LiamLimeLarm]

Suggestion: make the bars inbetween each element decrease in width horizontally instead of scale with size, i feel like that would make a lot more sense visually because of the problem with a bar that is 10% being nearly the same size as a bar that is 5%

[@DogsRNice]

I feel like a good way to show lots of values having similar sizes would be to color the region around them, so if there's a lot of similar values they'd be immediately obvious Or instead of color you could use brightness to make it more accessible for color blind users

[@cobble616]

Here's my thought to show visual ratios: Make the first place one have a border around it, then each entry has a border that is proportional to its ratio when compared to first place. Like bar graphs, just wrapped around the rectangles. Making them squares might make it easier to read too.

[@WhiteDragon103]

One visualization that I liked was bubbles that grow and shrink so that their surface area is proportional to the value represented. You could also combine multiple bubbles together when they get too small, and appropriately change the label. You could also make the bubbles try to push themselves into an ordering based on size, like small pebbles sinking to the bottom. Small bubbles belonging to the same category can seek one another out and merge into a single bubble representing that category.

[@johnd6980]

I think that the most efficient graph to show all of this data is somethong akin to a stock market heatmap. If you have seen one before its a list of many stocks, typically in the same sector. Each stock is represented by a square on the map with the size of the square representing its market cap, and colored red or green to signify the change in price over a time period.

[@ketchup9312]

Cary Ketchup Holder uploaded? What a surprise!

[@CTCtheHistorian]

Love to see a brand new way to look at data! Although, I will be sticking to the grid/table charts as that’s my personal preference, I do like what you’re doing here, especially with your decision to show the US states overtime. Also cool of you to reference subscriber count based videos again!

[@pWEC1kgW1X6c]

This is a really cool take on data race visualization! A nice addition would be to add some at-a-glance indication of the chain direction. Since each column alternates whether the larger items are above or below the smaller items, you still have to read the numbers to see this when comparing 2 items. Maybe having an arrow instead of a rectangle for the barriers would make this clearer.

[@GavinAnimates]

Welcome back Cary Kicks Harry

[@contrl31]

Great vis! You could even add a color (or asymmetric shape, like arrow ends) to the barriers to represent if the gap is growing or shrinking!

[@wack1305]

You made the most easily watchable tutorial I’ve ever seen. Content brained tutorial, but I understood every part

[@Asterism_Desmos]

Love this idea, very cool solution to a unique problem 👏

[@lekhakaananta5864]

Use a clock chart (I just coined this). Each contestant/item in the race is a hand on a clock. The leader is always at the top, midnight; its slice is a full circle, the whole pie. Everything else is scaled to an overlapping pie slice that starts from midnight and grows clockwise. This also solves the problems with the bar chart mentioned in the video. It scales in 2 dimensions. It saves space by overlapping the bars (after each bar is converted to a pie slice). It also keeps the advantages of a bar race without making compromises that the barrier chain does (sorry). It doesn't zigzag. It doesn't need additional barriers to represent deltas. It preserves proportions.

[@LiamLimeLarm]

thats pretty cool actually, its a bit hard to read at first but i get the jist

[@RolyPoIy]

the return of cary kite higher

[@yorgle]

Really clever and good visualization of the data! Somehow i wasn't subscribed to you before, even though i've enjoyed your videos in the past. Re-subbed!

[@TheSheep1]

I love when Cary Keeping Holes uploads in the holes that he keeps

[@spockfan2000]

That's brilliant. I am sure people will start using this - especially in science papers. Really amazing. Congratulations!

[@djdonny2]

i love when cary just rambles on about stuff idk about ^^

[@Lylcaruis]

lets go carykh upload ive missed these uploads

[@furbyfubar]

I agree with other comments that the barriers should show which side is bigger; probably by making them pointing in some way. Shrinking and growing < (or >) signs would make sense as it's a symbol already understood by most. But since you also want to show "What is the percentage different?" without needing calculations from the viewer, would writing that number on the (still growing and shrinking) less than or greater than sign make it too busy? If so, should it show the percentage of growth or shrinkage? Two adjacent elements have values 3000 and 1000, I think having the > say "200%" feels like it makes more sense than having it say "66.7%". In that example it probably shouldn't say "300%" as when two values move towards each other the greater than sign's size goes to 0, så having its label going towards 0%. Also, "3000 is 200% greater than 1000" and "3000 is 300% of 1000" are both true statements; but if we're using > as the symbol then it *should* be able to read it as "greater than" as well. Of course the < signs in the same space on your chain would always be pointing in the same direction as the snake doesn't change, but it would make it much clearer which values *are* closest to each other. Another small gripe: You keep saying it's "intuitive". But I doubt most viewers could tell you what was going on at 0:10 *before* you explained it. I certainly didn't notice the snake layout, so to me it looked like a bunch of states with numbers on them, placed in a grid. So I think making the snake/chain like structure more obvious visually is something you should focus on. If you skip the grid aspect and instead focus on the snake you can have the *distance* between the items in the graph *also* shrink and grown with the size of the < sign (or the barrier). If you also make it so that items can curve around the curves of the snake, then it would be much easier to see that the items are racing compared to each other. If you make the items circles (instead of rectangles) and have them move along a snake of constant width, two circles that are barely touching could imply they have almost the same value, and then still make the "full" swap once one overtakes another so that circles don't overlap (more than for the quick moment when a swap is happening). So it would then look like a snake that has swallowed a bunch of pool balls. That would give the advantage of having the distance between (the closest part of) two items being the same as the distance between those two bars in a bar chart. Sure, for the distance between two items that have another item in between them the distance wouldn't be their value's difference as the diameter of the circle in between them also has to be removed. But what you gain in being able to have the spacial dimension of the race brought back is probably well worth that wrinkle? I also agree with the comments saying that a spiral pattern might be more readable. As "clockwise" could then *always* mean "higher", as compared to now where "up" can mean either "higher" or "lower" depending on where on the chain you're looking.

[@Photosounder]

That's a good idea although not immediately intuitive. There's something else I do kind of by default when I do visualisation, I use colours to visualise a value. It could be a shade of grey where brighter means a bigger number, but if that won't do I use a formula when each primary colour (RGB) is determined by a raised sine like 0.5+0.5(freq*value+phase) and the frequency of each colour's sine is different (I almost always use 0.36, 0.187, 0.13 for red, green and blue, phase is 0.2 for all and I often add 1.5 to the value first so that it starts on blue and not grey) which gives a pleasant non-repeating smooth sequence of colours. So for something like state population I'd use the logarithm of the population, scale the value so that neighbouring values look somewhat different but not too different so you can see some continuity along the whole series, and so when two values would get close to each other you'd see that by how they become the same colour. I do this in my video player where there's a bar that represent the timestamp of already loaded frames in memory from the timestamp of the frame already loaded, so normally there's a smooth green to yellow gradient between the rectangles that represent each frame, but if the video has missing frames or timing discontinuities then you see something different, like abruptly switching to other colours at some point.

[@JonahWho]

yoooo cary kick hotdogs is back!!

[@Ido_Animations]

Just yesterday I wondered when are you coming back

[@Taco11037]

I LISTEN TO THESE VIDEOS LIKE PODCASTS. THANK YOU CARY!

[@Crawsome_Crustacean]

Barrier chain for every bfdi video and its views 🔥😤😤

[@tweeantelope784yt9]

This is the barrier chains video of all time.

[@Joel2Million]

great work as always, you already touched on the issue of 10% and 5% barriers looking pretty similar maybe just make the barriers 2d? make the thickness constant and have the length change?

[@GabeFrank]

a neat way to convey individual item growth/recession and distances between then might be to replace the barriers with collections of lines, like a topographic map. You could densely pack lines together between items, and have the items cross over individual liens as they meet/fail-to-meet those threshholds. You could even label or colour those lines they cross over with milestone quantities.

[@man-o-valor]

My only critique of the chain is it's a bit hard to tell which way the snake is going in the middle of a big chain. :^)

[@Alexifeu]

Very cool :D I quite liked it. I was sceptic at the beginning but in the end I really liked it.

[@MattSimmonsSysAdmin]

Really cool visualization! I would like to see the snake guides that are white set to black, and I would like the black barriers to be colored something like green/blue for growing/shrinking. I think that would help it stand out a little better.

[@arcturuslight_]

One visualisation I have seen that solves these problems is where you populate the screen with elements (usually circles) of different sizes based on the data. Similaely sized elements are clustered together, with largest element in the center. The drawback of that method is that it is easy to lose track of small elements, or even make them too small to see, which can be bad if you care about representing them.

[@MisterPenguin42]

Apologies for commenting again, this was an amazing vid.

[@Balloon.74]

Yo I saw you on that BFDIA 15 voting results live stream, glad you acknowledged my existence as a human with autism

[@Thegoober80085]

You kind of sound and look like my dentist 😰

[@Taco11037]

What if he is 😨

[@blabbilizer]

You just reminded me that my ex said I sound like Cary

[@Thegoober80085]

@@Taco11037 IM SCARED NOW HAS CARY SEEN ME CRY AFTER GETTING MY BRACES OFF 😭

[@Taco11037]

@@Thegoober80085 Maybe ¯\_(ツ)_/¯

[@TheSheep1]

Cary Kleans Hteeth

[@somdudewillson]

Pretty sure you generally want to avoid using areas like that, as it is well-studied that humans are significantly worse at intuitively comparing areas vs, say, lengths. I think this would work better if it was just the length of the barriers that scaled according to proportional difference.

[@DeathEaterLink]

Hey Cary keg holder, I think this idea is a good premise, but I don't fully agree with the "intuitiveness" of the barrier chain vs the bar graph. I think visually it is much easier to tell at 6:36 what information is being displayed on the left side. edit: nvm, you basically exactly addressed my point at 8:11. I'm sorry to say though, but I think the "visual ratios" are more important than the "n^2 scale"

[@CCABPSacsach]

4:56 Ohoho, I would like to see a DevianArt/Reddit post showing all these states in a map

[@deamichaelis1]

Attach a violin-type graph to the gray background snake and you get the ratio from one end to the other better. Then this would be a godly chart. 🥇 Just make the max at the high-end scale to the max width of the snake. You could just layer it on top of the gray background but behind everything else so you still get that clear path to follow. Make it white so it doesn't clash with any of the other colors. Or just let all the colors be customizable.

[@tricolorcircle]

"well now nobody fulfills everything" I propose the pie chart race. It's like a bar chart race but a pie chart

[@liammyt]

liamm, a new carykh video is out

[@chugz_for3v3r]

This is awesome cary! Thank you so much! Keep it up! You are awesome!

[@LUVVEOUS]

I have this idea of representing quantity as a percentage of transition from one color to another. For example, the one on the top of the top would be green, and the one at the bottom would be red. And the ones in between would have the color in between them corresponding to the percentage of value between the top and bottom

[@SansULB29]

*And he returns!!*

[@Hexagonal_Goblin]

This rocks! like you say, it's not the perfect catch-all, but it absolutely fills a useful data-vis niche.

[@whisperSSG8]

Someone has probably already mentioned this but sometimes you just want to know how big something is compared to something else not just something directly next to it (like the top vs #10) to get that sense of scale. I think a colour map gradient of every value would solve this as it makes a "third" dimension. This will also solve the issue of identifying whether the "snake" is going up or down in any column. For your map there will be 2 colours if you also want to show the US state groups.

[@maxneevo]

I'd like to see how a heat-map for a grid of numbers with possibly gradiants would rank on the ven diagram you put for the different methods

[@sugrado]

bro is sticking to Processing, gotta love the commitment

[@robertdixon2555]

One thing you can do is first off have the icons of each block has a color associated with a certain size and then also have a 'history' of the size of the barrier. Make the barrier black but then maybe the largest that barrier was between those two competing blocks in that specific position can be displayed as a different color bar behind it. This then also gives a sense of 'how big was this gap and how much has been overcome'

[@theblurjay2986]

Hey Cary! I was just going to comment to say I definitely remember there being data vis races like this where they used bubbles to display the largeness of a value (population in this case) that could grow and shrink. While these types of data vis do have wasted space and can sometimes be difficult to read, I think they would satisfy your two requirements and the third of being able to tell the difference between them. You could even update this data vis to display population as a percentage of the total, so you could switch the bubbles to rectangles and have the rectangles start by taking up the whole area. Or, you could set one rectangle (the most populous) to stay the same size like in the moving bar graphs! Idk, just an idea. Hopefully I explained this well enough 💀

[@scottinator1074]

If you wanted to satisfy all 3 requirements, I think you could turn the bar graph race into a race on a inward or outward spiral, where each entry is a contestant in the race. This would scale with n^2, as a spiral fills the whole screen, would easily visualize gaps between "contestants" and could provide a sense of proportionality to the visual. Markers could even be made along the spiral to more easily indicate milestone numbers, allowing for easier estimations of proportions.

[@pastamanofficial]

yay cary uploaded!!!!

[@kuzeyrl]

This is cool, the barrier size might be confusing at first, but still cool.

[@gudadada]

Babe wake up carykh posted

[@AstroDaTV_official]

Cardiac kiloton is back!

[@Lukepuke311]

a dynamic pie chart could work because it scales well, is intuitive and the ratio can be seen

[@alandrome7886]

There's a lot of comments so this might've been suggested before, but I have an idea on how to show relative size of the elements. The background of the snake (behind the pictures) should be a black-white gradient, where the element with the largest data point will have a black background and the element with the smallest data point will have a white background. This could show relative size as you could see how steep the gradient changes along the graph.

[@tycho25]

I'd replace the barriers with bars that extend left to right so it's easier to compare their sizes, but otherwise this is neat and I expect to see people using it soon

[@user-vw4xp5nt9f]

the barrier chain IS pretty cool when you *do* get used to it though :D