
Accessibility NYC hosted a Meetup last year at thoughtbot, titled “I Am Colorblind, and You Can Too!” The presenter was Jake Voytko, a senior software engineer at Etsy. Voytko addressed how people see color, how colorblind people are different, and how we can design around that difference.
Here are three moments we found especially interesting in the presentation:
- Does this sushi look tasty?
- Playing Candy Crush colorblind
- Twitter bot reveals how your pictures really look
Transcript: I Am Colorblind, and You Can Too!
Hello, everyone! Welcome to the accessibility and inclusive design meetup. I'm Cameron. This is Thomas and Shawn. They're the co-organizers of the meetup. We meet the first Tuesday of every month to talk about accessibility in a broad range of disciplines. And today, we're here at Thoughtbot. I want to thank them for hosting us today and for sponsoring the food, and also to thank SSB Bart Group for the ongoing sponsorship of the meetup. Also, Joly is here. We're live streaming today. So you may get some questions coming in from the world wide web. And without further ado, I'll hand it off to Jake Voytko. And he's gonna talk to us about colorblindness. (applause)
Hello. My name is -- as he said -- is Jake Voytko, and welcome to my presentation. I am colorblind, and you can too. Where I explain how eyes work. Why my eyes don't work. And kind of the ways that you can design around that, knowing the principles behind it. So why am I here at all? I happened to -- I used to work with Shawn Lauriat on the Google Docs team. And he saw my talk. I gave it a couple months ago at another friend's meetup. He asked me to give it and modify it for you guys, so I'm here. If you're interested in contacting me after this, I'm extremely easy to find online. JakeVoytko on Gmail, Github -- any service. The last time I gave the talk, I noticed the color was off on the projector.
It looks okay here, as far as my eyes can tell. But if you want to look at the talk on your phones, I have a Twitter bot set up that has a bunch of afterpictures that shows what the world looks like to somebody who is colorblind. So alternatively, if you want to send a picture of your own, you can tweet it, @Jakewouldsee, and it would respond with a colorblind version of your image. I wrote it in NodeJS, so it will probably respond.
So this talk is divided into three pieces. First, as I mentioned, I'm gonna talk about how do eyes work normally, if you have healthy eyes with full color vision. I'm red blind. How does the world look different to me? And why does the world look different to me? Then I'm gonna tell you kind of how to model it. There's a very easy way to model this, in color vision processing, where you can get pretty good estimates of what I see. It's not perfect, but it's a pretty good estimate to give you an idea of the challenges that I face or don't face, rather, in my day-to-day life. And finally, I'm gonna talk to you about some simple techniques that you can apply to your daily life, in working as developers
and designers, to avoid the issues here. So it's kind of a motivating example -- the format for all of these images is the same throughout the talk. Where the before is on the left and the after is on the right. And these doors are doors that are at Google. And the idea is that they're cafeteria doors, and they're very big and heavy, and people kept swinging them into people and hurting them. So the solution that facilities came up with is that they would color one of the doors red and color one of the doors green and no one would want to go through the red door and that would be fixed. I learned they did this when somebody yelled at me for going through the red door. So that was an interesting way to learn about
that. That just goes to show that you can't always strictly speaking design things based purely on color. So let's get into it. How do eyes that are fully functioning normally work? You detect color in your eyes through these three structures that are called cones.
And the three different types detect different parts of the visible wavelength. You have long cones, which I myself am missing. Which detects the ones listed here. Reds, yellows, oranges, and greens. Then you have the medium length cones that detect a lot of yellows and greens, and on the other side of the spectrum, you have these cones called short cones that detect blues and cyans. Interestingly, if you took the lenses out of your eyes, the short cones could detect ultraviolet light, but your lenses filter it out, so your eyes can't see that. And what your brain does, for any part of your vision -- it takes all of the cones in that part of the vision and combines them together. So in one part of the vision, your long and medium cones are saying hey, I'm getting a big hit. And the
short cones see nothing, your brain is gonna go -- oh, this is yellow or yellowish. It's also useful to know, because I'm producing these images for computer monitors, how do we represent color in the real world? Obviously computer monitors can't just put, like, a real color somewhere. It has to kind of fake it somehow. So it's useful to know for people who are new to... How images work how color is represented. So the idea is that instead of, like, representing the full spectrum of colors, you have three different colors that are called primaries. And each of the primaries targets one of the three cones that are in your eyes. So one of the primaries targets the long wavelength, one targets the medium, and one targets the short wavelength cones. And by modifying how much one of those primaries
are, you will modify how much that particular cone is responding. There's some overlap between the cones, so it's not perfect, but you can do a pretty good job of picking colors, so you can represent most of what you can see. Something interesting that I learned when I was learning about all of this is that you can't actually reproduce all colors using just three primaries. Like, it's not possible. But you can do a pretty good job. It gets most of the job done. And in grade school, I had learned that primary colors can reproduce all colors, so that was kind of an interesting fact that I had learned. In additive color spaces, which are in color spaces used on, like, computer monitors and projectors, where you're adding together lights to produce other lights, you are using RGB. Red, green, and
blue. And one version in particular is the one that is, like, the predominant one, on phones and computer monitors, and so on and so forth. And on the bottom here, I show you those three colors, which are red, green, and the landmark Miles Davis album, Kind of Blue. So what's different about me? As I mentioned a few times already, I am completely missing the long wavelength cone. This affects about one in 200 people. It's kind of a rarer version of colorblindness. About 5% of the population are colorblind. And the most common type is to not be missing a cone. But to have kind of like a weakened version of one. So that the other two cones dominate the third cone. The effect of this on me is that I really
can't see reds at all. I kind of see them as very dark. And oranges, yellows, and greens are affected too, to varying degrees. To give you an idea of how different vision is for me, most people can see about 150 different wavelengths of color. In your head right now, guess how many different wavelengths you think I see. Shawn knows the answer already because I sent him the slides. But the answer is I can differentiate about 17 wavelengths. So a little over a tenth of what most people can see. That's not to say bright and dark green. That's just if you look at a pure spectrum, that a prism splits, of pure white light.
So the... As I mentioned before, there are... Yeah, right? As I mentioned before, the most common type of colorblindness is to have two cones that dominate a third cone. So using some clever filtering techniques, you can attenuate that response. And there's a company called Enchroma that makes these glasses. You may have seen videos online of people putting them on and crying at sunsets. So it takes those two cones and attenuates them so they're just as strong as the third cone. So you get these balances of colors that you've never perceived before. You've never been able to see it before. So people with these weakened conditions can perceive colors. I was excited when I heard about that. I went online and did their vision test. Sorry, you're too colorblind for this test. But as a consolation
prize, here's a whole bunch of information about your condition. So that's great for me. I love reading about, you know, colorblindness and learning about things in general, and reading through it. And one sentence really stuck out to me. People with protanopia, which is the condition I have, will even perceive colors in the wrong part of the spectrum.
For instance, they'll perceive peanut butter as green. I needed to read that sentence three times before it really sank in. Because my entire life, I have thought that peanut butter is green. I found out last year that it's in fact not green. It's kind of brownish, which is news to me and no one else. So now that we know what's different about me, it's useful to talk about kind of how to model this, so that we can program it, so we can produce an image output. So there's a lot of my chicken scratch on this slide. I'm gonna spend some time here, because it's a little bit dense. So when you are working with colors, it's not really convenient to work with RGB, because it's a three-dimensional space. It's more convenient to work in color spaces that are designed around how our eyes function.
There's one in particular called XYY, where it takes brightness and makes brightness an entire axis. So if you only want to work with color, you can throw the brightness axis out and work only with what's called the chromanence axis. X and Y. So it's luminance and chromanence.
So the way colors fall in the spectrum -- all the primaries fall along this upside down U thing. Red, orange, yellow, green, and an eye that's functioning fully will see anything within this U.
Do you mind pointing from the other side?
You don't like reading text on my face? So... When I was mentioning before that there are three primary colors, red, green, and blue, they end up as this triangle inside of this color space. Where red is about here. Green is about here. And blue is about here.
If you pick other primaries, it'll form, like, a slightly different triangle. But the idea is about the same. And then white ends up all the way in the center. So the difference between what you see and what I see is actually a little obscured by the teleprompter. There's enough here that I can talk you through it. So what I see differently is instead of seeing everything inside of this U -- I only see a single -- oh, thank you. I only see a single parabola that intersects the U. So that gives you another idea of, like, it wasn't until I found that out that I realized how compromised my color vision is. Everybody sees the world in these two dimensions of color, and for me, it's collapsed down into this one line.
And for anybody who is monochromatic, they only have one cone, it would only be a single dot somewhere, around here or around here, probably. That was a big wake-up call for me. But that makes it very useful for modeling relatively accurately what I see as a colorblind person. Because all you need to do is find a way to transform this two-dimensional graph of colors down into this parabola that I see. So a lot of... You know, vision researchers have done a lot of work to determine what colors I confuse together. And it turns out that they mostly form these whole linear branches, where any two colors that I confuse will fall along these lines, which you see radiating out here. And also extremely conveniently,
they all tend to meet at a point. So the consequence of that is if you draw... If you have a color, a real color, that falls somewhere inside of this U thing -- I'm shaking the projector. I'm sorry. All you need to do is take this point, and take the point where the real color is, and find out where that line intersects this parabola. And it's not 100% perfect.
I'll kind of go through where -- what the strengths and weaknesses of that approach are later, but it does a very good job. I'm sorry for having to take you through my chicken scratch. I didn't find an image that kind of portrayed this quite the way I wanted, so I kind of hacked together my own. So I programmed this the first time. And it looked pretty good, except that there were some brightness issues. It seemed like the hue was right, but the brightness of the image was completely wrong. All the papers that were basing my implementation on were all back from the same paper in 1974. It was available for free online, thankfully, and it was a massive, dense paper, 30-plus pages of dense academic literature, and I read through it and understood it the best I could, and at the very bottom, in a
footnote, there was a mention that people with protanopia, which is the kind of colorblindness that I have, they perceive reds as being very dark. You can't take the luminance of a normal person and assume that's what I see. You actually need to model it separately. And that actually made a lot of sense to me. People -- when they find out I'm colorblind, what color is this, what color is this -- when it got to things that are red and green, I could generally tell. If something was dark -- stoplights and emergency lights look red to me, because most people don't color things dark green. People were like -- you're not really colorblind.
I promise you, I am. So there was a simple formula that this paper had, that let you model brightness, and once you applied it, the results looked pretty good. There are still some issues which I'm gonna talk through. But it's a great first approximation of how a colorblind person sees the world. And I'm talking a lot. If anyone has a question, just raise your hand, and we can get to you. So, again, I have a Twitterbot. I'm not gonna talk too much about this. Except to say that it was able to engage my mother in my colorblindness.
Whenever she sees a painting or something like that, now she sends tweets at my Twitterbot to see what her brother or cousin or me would see. So her entire life, she's been surrounded by colorblind people, and now she's finally able to get an idea of what I see as a colorblind person. I went through all of the images that I've ever taken, any picture I've ever taken, and I calculated what was called the root mean square deviation of all of the images.
And basically what that means is -- instead of summing together all of the errors, basically the original pixel versus the new pixel, I summed together the square of the errors, with the goal of finding images that had regions that were very different. And out of all of the images that I've ever taken, this was the one that was the most changed before and after. I do see a little bit of a color difference on this projector, so it might be worth bringing up on a phone. But the before and after... Look basically identical to me on a phone.
And that's kind of ironic, for a lot of reasons. Primarily of which -- I painted this myself, at a bring your own beer painting class. So it just goes to show that with enough careful instruction, I can accomplish just about anything. So when I was making this, you know... Question?
I'm sorry, could you say that again?
The head lines at the top of the slides are deep red. And she was asking how do you know that they're deep red.
I can read the hex values of colors. And tell that way. This is a slide deck that I did through Apple. A lot of times I rely on default presentations and default styling of things. So any styling in this presentation is all thanks to Apple, and any errors are thanks to me. So when I was implementing this, you know, when I do a good job implementing this program, the image that I put into it looks the same to me as the image that comes out of it. So I was bugging my friends endlessly, sending them before and after images. What does this look like? What does this look like? So for this flamingo picture, I sent it to my friend and asked if they were different. She said they were very different. She said
the flamingo and the sun, they were very orange, and pink. And the second was this dull grey and yellow. That made me very happy, because my program is working now. And she responded to me with frowny face, your life is sad. Which, since I made this program, is a response I've gotten a lot. No, it's not. And another thing I was really excited to try was... Finding out what color the dress was, or more accurately, testing my friends. For those of you who don't know, there was this phenomenon that went around the internet last year, where this British teenager was shopping for wedding dresses, snapped a few crappy cell phone pictures, and posted them on Instagram, because her and her friends all thought that the dresses had a very obvious color, but half of her friends thought the dress was obviously white
and gold and the other half thought it was obviously blue and black. And basically it turns out that it depends on whether you think it's in a dark room or a bright room. It causes you to think one or the other. I've only ever been able to see it as blue and black. So once I had this program, I was like -- hey, I have friends who can force themselves to see it both ways. Maybe I can only see it in blue and black because I'm colorblind.
I ran it through, sent it to my friends. No, I can definitely still see white and gold in it. The problem is on your end. You can't just blame it on your eyes. That was a sad day for me. I can't just see it as white and gold.
Awwww!
I get that a lot. So this is another interesting image. When I was preparing the last talk that I gave... Do you want some? So when I was preparing this talk the last time, I put these up, just because it was one of the top most different -- from before and after images.
And my friend Sarah, who ran the Meetup that I was presenting at, made sure I didn't swap any images or whatever, asked me -- you know, was I out of my mind when I ate sushi? She was like... The one on the right is rancid. I would never eat it. And asked me a very interesting question. Does sushi actually look appetizing to you? And truthfully, no.
It does not. The salmon in the center, yes. That looks tasty. I would eat that. But whatever is on the top or the bottom, if I didn't know that my friend was not trying to poison me, I would never put it in my mouth. So let's get into some of the problems with the approach that I have. So this was an image that is very interesting. It is, I think, much more different to me than it is to other people. So I see the table on the right as being maybe like twice the brightness, 50% to 100% of the brightness as the table on the left. Most people who look at this say they don't appear that different. The difference isn't that much for brightness. So what I think is happening is that... The processing technique that I
use shifted the red, and didn't... The linear correction for brightness doesn't do a great job for me in particular. So it comes out a relatively close before and after for brightness for somebody with regular vision, but it doesn't do it great for me. This was actually... Again, I had a friend look over my last slide deck and she was saying that I should probably take this image out altogether, because it didn't look very different before and after.
But I'm leaving it in, because the difference is actually pretty significant before and after for me. But not necessarily for other people. These kind of shifts happen in other images. So there are two errors that I can detect in this image. One is I have the same type of brightness image in the before and after on my shirt, and on my dad's shirt, I see a huge shift. It shifts more towards green on the right hand side than it does on the left hand side. To have another image where this hue shift happened, this was probably the image that appeared the most different before and after to me. Just because the table -- I can see a noticeable shift towards green. From what I imagine is red. So I sent it to
my friend Hanna, who is in the picture, and I said -- hey, can I use this picture, whatever, and she was telling me that the whole image looked different to her. For me, the only difference is the table. She was saying everything is different, down to the color of her skin and stuff like that. So while, to me, this seems like a failure image, I think it still does a decent job of kind of portraying the before and after difference for people with my type of vision problems, protanopia. So let's get into the final part of this talk.
Kind of what can you do. So as kind of a motivating example, here is the colorblind processed version of the Late Show with Stephen Colbert logo. So with normal vision, I'm sure this looks totally fine. You can actually see reds as having a brightness and vibrance. So it would be bright red on blue. But for me, reds all appear very dark. So the Late Night with Stephen Colbert logo has appeared very strange to me. The rest of the typography is very crisp and pops out, but you have the Late Show which hides, which is this dark red on dark blue. So I hope this comes through for you. How do you avoid doing things like that?
For a lot of sites and designs, if you're working for small audiences, it probably doesn't matter, but I'm sure the designers of Late Show with Stephen Colbert, who are trying to appeal to every person in America, they would probably want to know that over a million people in America think your logo looks kind of weird. So there are a couple of techniques --
A couple questions. Is it don't put red on blue, or if you do, one better be a whole lot darker than the other?
It kind of hurts to look at, so please don't in general. But to your point, yes. It would have to be... Very, very, very bright red on a very dark blue. Yeah, exactly.
Just to follow up with that question... I don't know how other more full range vision people feel in this room... But I find from a design standpoint, red on a solid blue background is a little... As you described... A little painful. It's jarring.
Well, it's very jarring for me. So I'm glad to know that other people feel my pain. There was another question over here?
Yeah, is that specific to your specific type of colorblindness? Or is that like... If somebody had... No medium cone, would that be true?
I'm not sure for the no medium or weak medium or no short or weak short. But for most the both common type of colorblindness and for my particular type of colorblindness, it's a problem. Where the most common type is kind of like the halfway point between what I see and what a normal person sees. Where you have, like, a weak red detecting cone. It affects them, but not nearly to the extent that it does for me. I have a good friend of mine -- has is protanomalous, which is the halfway version of protanopia. And he says it's a problem, but I complain about it more than he does. So I think it affects me more than it affects him. So let's walk through the techniques, just one at a time.
They're kind of very simple and very easy to apply. And I'm just gonna show you -- sometimes applications that do these things very well, and sometimes applications do these things very poorly. So I'm a recovering Candy Crush addict. I never spent any money on it, but I spent a lot of time on it. Even though it's a very color-based game and a very colorful game, every color is paired with a unique shape. So they don't use just color to differentiate any single thing in their game. They just use it as a highlighting thing. The problem I would have had -- there was like a hard candy and a jelly. They look very similar to me. This is the colorblind processed version of my image. And I was able to play the game, because they gave all of the candies such unique shapes, and different brightnesses
and stuff like that. That really, with any type of vision impairment, just so long as you can see at all, you'll be able to play the game, because everything will really stand out to you. There's something later that's a follow-up. I kind of ordered these slides a little poorly. So another thing you can do is test in grey scale. Can you do me a favor and flip the -- I didn't realize there was gonna be text on the bottom. So I would have designed the slides around that. So one thing I did when I was trying to make this slide is -- I went on the Data is Beautiful subReddit, and I scanned down until I ran into a visualization I found problems with. And it didn't take very long to find one.
Someone had done a sentiment analysis of a bunch of different subreddits that are sports related for when they're happy and sad. And the problem that I had, which now you all have, because I made this image grey scale, was that they made sadness red and happiness green. So when things were at extremes, they looked kind of the same to me. So when I looked at this thing, I wasn't 100% sure what the graph was actually representing, because it's labeled moody subreddits. I didn't know if they were very moody if they were unhappy or maximum was positive because they had a lot of mood. So I don't know. One thing that the visualizer of this data may have wanted to do would be to flip it in grey scale. As
soon as you look at that, it doesn't matter how much color vision you have. This is a problem. It's really hard to tell what's going on here. So they may have tried to find another method of representing this data. So the good thing about testing in grey scale is that you don't even need to worry about what kind of colorblindness. If it works in grey scale, it will even work for people with monochromatic vision.
You said reds are much darker for you. I know you can turn on greyscale through the accessibility panel. Will that capture that experience that you have with the darkness?
It's hard to say, because so many programs implement greyscale poorly. It turns out that different color channels contribute to brightness in different ways. And most people don't know that. So the greyscale representation is usually not very accurate. So it's kind of hard to answer that, generally, for all things.
We can do a test afterwards. Basically bring up one of the slides with the side by side, and put your Mac in greyscale, and we'll see what the differences are. Question over there.
Yeah, on the point of going greyscale, if you were to invert the colors, sometimes like a high contrast mode does that... Does that affect the perception of the differentiation between red and green for you? Since the red at the top here...
That's another interesting point, where it's hard to generalize that. Because the luminance of greyscale is not linear. It's actually done with something called a gamma curve. Where the medium grey, the neutral grey, wouldn't be colored 128 and 256. It would be... I think it's 186, actually. Most programs don't know it's not linear, so it's one of those things where... In practice, it should be fine flipping it. It should give you the difference. But... Because of most people don't implement it the correct way, it's not... It's likely not going to work. And for different reasons. Yeah, this is one of those things where knowing this much about vision processing turns into more of a curse than a blessing. Please underline your links, oh, god, oh, god, please underline your links.
My entire webfaring life, whenever designers sway away from the blue link on white background palette, I have so many problems. I can't tell you how many times -- I've read blogs for years and happened to mouse over a paragraph and suddenly my mouse cursor flips into the -- please click this link cursor. And I never realized that the blog that I've been reading for years has been linking to dozens of different sources that I may have found interesting or insightful. There's a good example here -- the colorblind processed version of just a story that popped up on my Twitter feed, where there are three different links. There are two @ mentions that are links, and there's a third, which is obviously a URL, which is clickable, but because it's done in red, I have a difficult time realizing that these
are actually clickable. With the context where... You know, where these things are obviously links, it's a little better, because just as an experienced web user, I know I can click these things and probably get where I'm going. But because this literally affects you from birth, someone who's not fluent in web trends or convention might have problems. They might not realize they can click it.
Might not realize they can go places. So this goes back to the Candy Crush example. Where it's very useful to pair together callouts that something is an action along with the color that indicates that it's an action. One thing that's not an underline, that I've seen some websites do is they include a little arrow next to a link. A lot of news websites do this. Where they show that you're gonna leave. There's an external link somewhere.
So while it's not strictly underlined, it's still useful. It still gives you visually useful information to say -- hey, this is something that I'm explicitly calling you to, that you can click and do something with. Oh, God, please underline your links. So this kind of brings us to the end of my talk. Or everything that I've prepared. To kind of summarize everything that I've gone over... Most people can see color normally. Because they have three cones in their eyes. I am missing one of them. So I have a reduced perspective on the world. I see a little over a 10th of the colors that most people can perceive.
It's very easy to model this, if you remember the graph that I had with the upside down U. Everybody can see basically in two dimensions on that graph. And for me, it's collapsed down into a single parabola. And by intersecting those confusion lines with the parabola, it gives you an idea of the colors I see. I wrote a Twitterbot, and I gave you tips to make it accessible, and oh, God, please underline your links. Please, please.
(applause)
Does anybody have any questions?
I have a question. So at the beginning, talking about 17 wavelengths... Is that another way to think about... Your vocabulary for describing color as you perceive it would be like 17 colors?
Yeah, that would be an interesting way of saying that.
So is that something that from a descriptive -- like, any other colors, other than red --
I see blues very well. I've always liked blue Christmas lights. I can pick them out very easily. I think I would confuse them with purple Christmas lights, because they have red in them and I don't see the red part of it. Yellows, bright greens, those are all colors I can imagine pretty easily. On the other hand, there are these colors in the middle that are, I think, the combination of a lot of wavelengths, like maybe (inaudible) and stuff like that -- I look at it, and I literally don't understand what I'm seeing.
Not only do I not know what it is, but it's hard for me to pin it to any other part of the spectrum. Yes, in the back?
(inaudible) colors that are more in contrast and also have shapes in them, so you can differentiate them better. And I really like that. The options (inaudible) and second is a Mac app called Oracle, that you can switch all the different colorblindnesses to see the foes that you're looking at.
That's a good point. Yeah. There are also ways that you can test for individual colorblindnesses. It's worth when you use them finding people who have these colorblindnesses and running it by them. I needed to do a lot of tweaking to my program to get it accurate. Like, I kind of went over the high level. But there was a lot of stuff. Apparently between different color spaces, they literally have different whites. You need to actually convert the whites that you see between them. There's a lot of processing steps that made it -- so the only reason that I got even close is because I already have this colorblindness. If I were programming this for somebody with deuteranopia, where they were missing the medium length cone, I would have to track down somebody with that condition and say -- hey, is this
close to what you see? So yeah, it's useful to use these tools. But if you're really relying on them, it's worth having a real person as a backup, just to verify that the tool is doing what it says it's doing. Yes?
Can you talk about... Like, are there really popular tools that you're referring to?
The woman in the back was just talking about them. You told me it was called Oracle? But I never use it. I would try to find somebody with colorblindness just to kind of double check what it was.
Fujitsu has one called Color Doctor also.
Yes?
Some people who have colorblindness don't tend to like the word colorblindness. Because it's not correct. Not a correct description. Because it's not that you can't see any colors.
Preferring the word... Color deficiency.
Yes, that's right. The technical term for it is red-green color deficiency. The term that used to be used is red blindness and green blindness, and they're extremely inaccurate. While I can't see red, people who are green blind can see green. So it's an old term from the '30s that's kind of persisted until today.
I just wanted to say I think your sushi demo is the most compelling colorblindness example I've ever seen.
That's why I need a taster for all of my food before I eat it. In the back?
Yeah. One thing I noticed when you were talking about the Candy Crush game, anecdotally, I would say that having a clear distinction aside from colors is useful for everyone. Do you know of any usability studies or research that points to that being broadly true?
I just realized I've been bad at repeating questions. So the question is: Are there any studies that have shown that pairing colors with shapes or labels is helpful? I've only seen anecdotal studies. There's this guy who's a really big design guy called Don Morman.
He wrote one of the handbooks on it called the Design of Everyday Things. And I've seen him give talks where he talks about this. I feel kind of embarrassed, because in Google Docs, we had just images. He says just images or just words or just colors are okay-ish, but the best thing to do is to pair them together. In one talk, he explicitly said the best way you can get a user to understand where you're going is to pair a shape with a label. And I'm sure color would even only strengthen that.
Is there any case of it having a negative impact?
I'm sure.
I can't think of anything.
Is there a case of having a negative impact? I'm sure it's possible to come up with it. But off the top of my head, I can't think of anything. It seems like one of those things where it's probably almost always strictly better. The negative impact is that you're taking up more space. Which is likely why most applications don't do it. Especially on -- as screens get smaller and smaller, vertical space is at such a premium that it's hard to justify it. If you look at Microsoft Office, they did a redesign of their ribbon, where if you look at the first screen, it's all of the images that you've already memorized.
Bold, italic, underline. They don't pair it with anything. But if you go into literally any other menu, they've decided the vertical space is worth giving up, to pair a label for all the images they have, for literally every other function Microsoft Office does. Yes?
I want to say it's really cool that you brought up the blue and red not being put together. I used to teach art for a really long time, and I would always see the desire to take every primary color at the highest chroma, just smash them next to each other, and one of the reasons that it doesn't work is because our eyes are terrible and fuzzy and really just not good at figuring out what's next to each other. So it'll actually start kind of crossing the signal and cancel each other out. So there's that type of vision -- right at those edges, it'll make whatever happens there darker. It'll create this illusion that things are actually kind of... Have this weird... Kind of the way blinds look like.
That's cool. I've never heard of that.
It's good for everybody too. It's good across the board to not combine blue and red.
Do you happen to know the name of that interference pattern?
It's one Google search away for me. But it's basically because your eyes are saying -- this is really red! This is really blue! And right where they're next to each other, it's a little fuzzy signal. When colors makes the -- that's why you can't mix a perfect color palette from the primary ones. Because they start canceling each other out, becoming more neutral. That's why when you mix them all together, you just get grey. It's that same thing. They start canceling each other out right there at those intersections, and our eyes aren't pixels. They're way worse.
Cool!
I just thought it was cool. It just never connected for this, for me.
Any other questions? Two more. Over here. You haven't spoken yet.
Hi. I'm a teacher. And I'm wondering as a kid growing up how your colorblindness affected your learning. Because I make posters and I use red and blue all the time. It's my go-to.
And I have the most amount of markers for it. So what would you have wished that your teacher had done to make things less aggravating for you?
So growing up, how did this kind of affect my education? And I guess the answer is that I may in some ways have happened to have lucked out. I'm a very analytic learner. I'm very comfortable when I'm looking at math equations and stuff like that. So I think maybe the color didn't matter so much to me. Anything that my teachers would do with color would be to try to differentiate two concepts with color to kind of highlight it. But that was always a very easy separation for me to do naturally. Split things up and categorize them and stuff like that. Maybe it would have strengthened some lessons, but I don't think it would have detracted from any lessons. Obviously art class was difficult. And there was one more question over here.
I was kidding around. I wanted to talk about red and green. Those are often used together because they're associated with Christmas. Is that... I mean, to me, that season -- I go -- uh-oh. Here comes the red and green. It's an assault.
For me it's here comes the grey and more grey.
So you got off the hook on that one.
I guess I did.
So if... For links, are you saying basically... Don't use red? Or in the red spectrum?
For me, specifically.
Commonly speaking, the red green is the most... Right? Isn't that the most common?
The default is blue. I don't know...
With colorblindness...
Yeah, red-green. That's right. Red-green is the most common type of colorblindness.
With so many major calls to action, and buttons being orange and red, and all of a sudden, they're looking darker for you, what would you suggest for this?
So calls to action are interesting, in that marketing people have kind of figured out that calls to action that are orange with blue color schemes are very clickable. That happens to be very easily differentiable for basically everybody. That's nice. It's really when you have custom websites made either by designers or for smaller audiences, where red is an easy example to make, but just because I have such a broad range of colors that can be confused together, it's very easy to find two colors that would look about the same to me, but would look different to somebody with normal color vision. So the pairing...
So I guess the rule would be either to stick with safe color combinations, or alternatively, somehow highlight that links are different.
So specifically -- if it's a contrasting color, you're saying it's a big difference. But if it's an analogous color scheme -- it's something where, say, somebody really wants the red and green color scheme or complementary... Or if they want that yellow and orange going... What would you recommend? And would it be messing with the luminance? Is there something where you're stuck with this color palette that you know is gonna be tough?
Underline the links, please. Flat out the easiest way to just get it working for everybody. Is to somehow call them out. Yes?
So when you're driving, the red is just a lot darker than the green?
I guess at traffic lights.
Okay.
No, I don't. I'm sorry. Old joke. That's a great question.
Probably a very common question, but I'm just curious.
This used to be a problem. They used to not design traffic lights to be colorblind safe. I drove through a city once that had non-colorblind safe traffic lights. Asbury Park had not updated their traffic lights. I'm driving through this light, and it's red, but the traffic light looks completely off to me. Nothing is lit up on it. I'm getting closer, I'm getting closer, is that traffic light red... My friend is like... Oh, god, yes! Please stop! So most traffic lights now are done in... It doesn't look like it's red, yellow, and green, but it looks very different to me. Like, the green looks very white to me. I don't see it as green at all. The red looks very dark to me, but I can see that it's lit up, and the yellow looks like mystery color. It might be orange, might be yellow.
So that was done with the colorblind community in mind. That was changed because of that. And it doesn't look changed, really, to us. But it does to you guys.
Yes.
That is very interesting.
My understanding is like in Romania, and possibly other countries, they have illegal... It's not legal to drive with colorblindness. That's something that... I guess globally, there's still an opportunity to evolve on that.
MIRABAI: I was just curious, because you mentioned that a lot of people in your family had the same type of colorblindness. But was your colorblindness discovered at a very young age? And do you think that's typical? Or was it mainly because your family new to sort of look for it and test you? Do you know what age colorblindness is typically discovered in kids?
That's interesting. So the question is... What age is colorblindness typically discovered and kind of how was it discovered for me? So my mom was the first of all of her siblings to have children, so I was kind of like the test dummy for this. They didn't discover it until I was in kindergarten. I went to a nurse and she held out a box of crayons and asked me to name all the crayons. And I couldn't name all of the crayons without seeing the labels. So the nurse called my mom and said... I think your son has a learning disability. And my mom asks why do you think that? The nurse explains it. And my mom says...
Colorblindness runs in my family. Did it occur to you that the test was for colorblindness? And the nurse was like... I guess that kind of does make sense. So that was when they learned it for me. I can't speak in general. In my family, once they learned that I was colorblind, the new generation was tested very early. So that when my cousin Jerry went to school, they already knew he was colorblind. It was something they figured out from the time you can teach a child colors. In the back.
Have you ever thought about using custom stylesheets? As in setting links in your browser to be a color you can see?
Have I thought about using custom stylesheets? The one way I can do that is using extensions like Readability, which strip away all the styling of web pages to present it in a readable manner. Strictly speaking, the issue with that is that I would need to do something for... Not just the text, but the background too. So it would stand out from the text.
So on one side, maybe I could toggle it on and off. But if I left it all the time, I feel like it would kind of ruin the web experience, to highlight it like that. But that is an interesting point. I've never actually considered that. That's an interesting thought.
I just wanted to say we had a few people watching the live stream today, which is our first time. Comment from the live stream, saying... From a Deborah Edwards, fun to hear the talk of red and blue given all the red and blue in today's political ads, news for Super Tuesday. So I guess another observation. I guess our political campaigners use that color combination as well.
But they don't mix. That's the idea. They never overlap.
And then everyone else is just saying -- thanks for a great presentation.
Happy to hear it. Thank you for watching.
Last call for questions. All right. Well, thank you, Jake. (applause)
So our Meetup is documented on Meetup.com. And also, we have a Twitter handle, @a11ynyc. Meetup is the same. Meetup.com/a11ynyc. Can you leave this plugged in just for a moment?
I just realized the captions are not capturing...
I'm sorry.
I saw Shawn furiously signing. Okay. So... I wanted to thank you again. Obviously thank you to Thoughtbot, thank you to SSB Bart. You're welcome to stick around for another half an hour or so. Get to know people, talk, there's some more pizza, grab a drink, and enjoy yourselves. Thanks again. (applause)