* I can't know your monitor's calibration, your ambient light, or your phone's brightness. Obviously, this will affect the results. However, I am tracking local time of day and device type, from which we should be able to infer whether night mode and default calibration has any aggregate effects. Anecdotally, thus far, I haven't found any effects of Android vs. iPhone (N=34,000).
* The order is randomized. Where you start from can influence the outcome, but methodologically it's better to randomize so the aggregate results average over starting point. You can run the test several times to see how reliable this is for you.
* It's common practice in psychophysics to use two alternatives rather than three (e.g. blue, green, something in the middle). It would be a fun extension, which you can handle with an ordered logistic regression. The code is open if you want to take a shot at it: https://github.com/patrickmineault/ismyblue
* I am aware of most of the limitations of this test. I have run psychophysics experiments in a lab on calibrated CRTs during my PhD in visual neuroscience. *This is just entertainment*. I did this project to see if I could make a fun webapp in Vue.js using Claude Sonnet, and later cursor, given that I am not highly proficient in modern webdev. A secondary point was to engage people in vision science and get them to talk and think about perception and language. I think it worked!
My partner and I regularly disagree on blue vs green as the colours become more of a gray colour - might be interesting to randomise the brightness of the colours being displayed then seeing if the skew towards people perceiving blue Vs green changes as the colours become closer to gray.
I also often disagree on blue vs purple, which is inconvenient when we name the same coat two different colors.
I think my "blue" is a way more specific shade than most people (hue 192 here, whatever that means on an uncalibrated display). Likewise, I'll usually say "purple" before others.
My partner and I were well aware of the limitations, but it has clearly demonstrated our difference in perceptions in a way we were both happy with. Being able to see where your partner lands relative to you is deeply satisfying.
It was fun but I messed up the statistics! I had Redshift running, which (maybe you know) makes the colors more reddish. And I got a bluer than 98% of the population result. Turning off Redshift ... makes me instead greener than bluer.
I would guess the hackernews crowd has a higher percent of bluefilter installs since that is a very common topic. Probably also more agressive settings for the blue filter.
When done on my Xperia cell phone, even a small shift in screen orientation made the green leaners into obviously blue. Might be worthwhile capturing phone position if you can.
The mask is 200 ms long, which is a bit on the long side compared to most psychophysics experiments. I can try to crank it up to 300 ms, but beyond that I think it'll start feeling slow.
The site records local time of day when you hit submit so I can track whether this has any effect. I have 7,000 answers thus far, I should have enough by tomorrow to determine whether there are any systematic effects.
Yes, you can absolutely get better at discriminating different colors, orientations, etc. though unfortunately improvements tend to be highly specific to the stimulus. There's a great book by Barbara Dosher called Perceptual Learning that extensively overviews the literature.
The order is randomized. Hit reset and you'll get a different sequence. The sequence is also adaptive (not a binary search---it's hitting specific points of the tail of a sigmoid in a logistic regression it's building as you go along). Try it a few times and you'll see how reproducible it is for you.
It of course depends on the calibration of your monitor. One of the reasons I did this project is I wanted to see if there were systematic differences in color names and balance in the wild, for example, by device type (desktop vs. Android vs. iPhone), time of day (night mode), country (Sapir-Whorf), etc.
The sequence itself should be converging however, right? I feel that there should be some random jumps outside of the current confidence interval so that contextual aspects can be filtered out or at least recognized.
Yes, exactly this. Because it seems to be converging right now, I quickly get the feeling that there's no meaningful choice, after the first three prompts you end up with something that's neither green nor blue. Re-taking the test gave me a very different score.
It might work better for me to do some contrastive questioning: show a definite green followed by an intermediary color, then a definite blue followed by an intermediate color.
The whole point of asserting where your border between green and blue is, is to ask about colors that are in between the two. It doesn't make sense to ask is RGB(0,0,255) blue to you? Well, unless you are color blind it is.
Of course, that's clear as day; the idea is to reset your presumptions from the previous trial and sample the ambiguous colors in a more consistent way, by priming you from the extreme ends of the green/blue scale.
It is common practice in psychometrics to use two levels in a forced choice and model responses as a logistic regression, which is what's done here. Adding an N/A option turns the thing into an ordered logistic regression with unknown levels, which is tricky to fit, but it's possible. Having done a lot of psychophysics, having more options generally doesn't make the task easier.
Sounds like psychometrics is unsuitable for modeling this problem, according to what you're saying. When you have a hammer everything looks like a nail.
The way that XKCD did it is the best, you ask people to give a name to each color then the responses are entirely natural and unprompted.
I don’t think that forced choice can give accurate results if a substantial number of people perceive green and blue as being non-adjacent - i.e. there exists a color between green and blue (turquoise/cyan/teal).
Otherwise it’s like asking people whether a color is red or yellow, when it’s clearly a shade of orange.
Are you sure that it is common practice for a problem that has three valid answers A, B and C, to only allow people to answer A or C?
Your website is not talking about "levels" of colour.
It's asking "is this blue or green", not "is this closer to blue or closer to green".
The question (1) "is this blue or green" has three valid answers: blue, green or neither.
The question (2) "is this closer to blue or green" only has two valid answers.
I would assume that with these types of surveys, the first thing to do is to qualify the proper categorization of the question.
Sorry to say, but to me it seems that almost all of the confusion in the discussion here is because you're asking question (1) (which has three valid answers) but expecting an answer from (2) (which indeed has two valid answers).
Author here, yes, it tests a mix of your monitor calibration and colour naming. The two types of inferences you can make with this are:
1. If two people take the test with the same device, in the same lighting (e.g. in the same room), their relative thresholds should be fairly stable.
2. If you average over large populations, you can estimate population thresholds, marginalizing over monitor calibrations.
The most interesting thing for me is that while cyan (#00ffff) is nominally halfway between blue and green, most people's thresholds, averaged over monitor calibrations, imply that cyan is classified as blue. I was not expecting that the median threshold (hue 174) would be so deep into the greens.
I got hue 174 as my threshold and really I just wanted to say "neither, this is turquoise/teal" for most of the questions. But blue/green was the only option.
I got hue 175. It's interesting to note that some older cultures, Japan for example, didn't always have separate words for blue and green, both were the same color ("ao" in Japanese). You can see the effects of this even today with things like traffic lights in Japan, which are considered "green" by their standards but blue by many others' standards.
There are also other cultures, such as Russia, where light blue / dark blue (simplification) are effectively considered separate colors.
All this to say, personally, I think we will continue to evolve to recognize more distinct "colors" such as teal, which is neither blue nor green but somewhere between. A lot of this recognition power is rooted in linguistics and culture, it's not as strictly biological as one might think.
Thanks for this comment! I dabble in fountain pens a bit, and one of my favorite inks is "ao" by Taccia.
Now it all makes sense (tho, to my eye it's kind of a blurple–royal blue; I get no green or teal from it. But, now I'm tempted to go do a blotter of it and look at it extra carefully in natural light.)
In Russian light blue is “blue” and dark blue is “indigo” essentially. It still has seven colors in the rainbow. It’s just that in English colloquially nobody uses indigo.
Yes, well that's what I mean. Culturally, Russians think and speak about colors differently, dividing them up differently than the West.
> Russian does not have a single word referring to the whole range of colors denoted by the English term "blue". Instead, it traditionally treats light blue (голубой, goluboy) as a separate color independent from plain or dark blue (синий, siniy), with all seven "basic" colors of the spectrum (red–orange–yellow–green–голубой/goluboy (sky blue, light azure, but does not equal cyan)–синий/siniy ("true" deep blue, like synthetic ultramarine)–violet) while in English the light blues like azure and cyan are considered mere shades of "blue" and not different colors.
> Blue: plava (indicates any blue) and modra; in the eastern speaking areas modra indicates dark blue, in some of the western areas it may indicate any blue
I am not deeply knowledgeable on Russian, I failed Russian in high school, just going off of my surface-level knowledge of linguistic relativity regarding color, and discussions with a friend from that part of the world, so I might not know what I'm talking about here.
That’s a bit complicated. The difference between синий and голубой is not really a difference in hue, it’s a difference in brightness.
It tends to be true that hues tending towards green are perceived more brightly than hues tending towards red, which means that blues with more green in them are more likely to be голубой, but by virtue of the fact they are perceived to be brighter.
But in principle, the line is drawn horizontally on the colour chart (or at least diagonally), not vertically.
The color name question here doesn't have a clear answer because most of the respondents would call this "teal", "blue–green", "turqoise", "cyan", "aqua", or some similar name. You'd get somewhat similar results asking whether an orange (the fruit) is really "red" or "yellow", or whether an eggplant is really "blue" or "red".
An individual person's answers on this kind of question are likely to vary from day to day, are context dependent (i.e. whether one object or another appears more "green" or "blue" depends on what kind of object it is), and colors this intense are very sensitive to changes in eye adaptation and technical details of the display and software, as well as inter-observer metamerism.
So in addition to the color naming difficulties, it's not even a very good test of color naming, if you want to get reliable psychometric/linguistic data.
For a single individual, all of the above is true, but for a large enough sample size, the answers may be more generally useful because you account for all of those rounding errors.
No, because if my case holds more genera (and I suspect it does), the answers are in part out of sheer frustration, and therefore prone to being similar to the last one given.
I didn't exactly rage quit but did think it was silly.
I wouldn't describe teal as blue or green any more than I'd describe purple as red or blue, so being forced to pick felt silly. Like being forced to choose my seventh favorite Norwegian glacier - technically its a valid question but my answer is necessarily going to be arbitrary.
That’s like asking which way a Necker cube is oriented. It’s both and neither. For blue and green, there’s a range of shades for which that ambiguity is true and you can “flip” it in your mind.
I would actually find it more practical to determine the thresholds on both sides where I find it to become ambiguous.
Not as far as I can tell. The phrasing of the question test does not acknowledge such ambiguity to start with, and by forcing them to answer one way or the other the test does not allow the users to signal perceived ambiguity even if they wanted to.
So how could the point of this exercise possibly be to find the range of ambiguity?
Fun, I got 174 and when I saw the results my reaction was "but that is not turquoise!" which I suppose means I either don't know what turquoise is, or my screen has bad calibration/gamut.
I don't think those specs make a difference. You would need a wide gamut display and a hardware calibrator to be sure you were looking at the colour as it should be
That wasn't clearly part of the test. To be ultra-pedantic (this is HN after all), the user's choices don't say "This is more-blue-than-green" and "This is more-green-than-blue". The choices are only "This is green" and "This is blue" forcing you to just pick one, where there is no clearly correct choice. When the color on the screen is neither green nor blue, many people will just pick a random answer.
I bet if the choices actually said "This is more green than blue" the results would be different.
On such a random internet doodad most users will pick a random answer period. To see what this thingy tries to do without wasting any time on it. I hope it doesn't try to do gather any meaningful data.
Personally I "tried" to answer truthfully at first and then went absolutely "ok f u, don't care no more" when it showed turquoise :D
Taking how you behave, and extrapolating that it to everyone, (and furthermore being unable to accept that other people might behave differently), is not a winning strategy for life.
According to conversion rates and engagement metrics of most apps I've seen (not even mentioning social media where 2-3% engagement is the norm) most users are ¯\_(ツ)_/¯.
Unless said app is a work/hobby tool, but that shouldn't be really called engagement.
Turqoise doesn't feel either more-green-than-blue or more-blue-than-green. It feels neither blue nor green, and I don't see any way to compare it to either.
It's clearly more turqoise than blue. Or green.
Turqoise on a computer monitor is always missing part of itself, so maybe I should've answered based on that, but I don't think the computer monitor was the point.
180 and blue and I suspect that language also plays a part (I was brought up in an environment where the word turquoise starts with green, but now live in a turquoise-producing state where the finished product look far blue-r.)
it looks like my default is if there is 40% green in that it is green. Thus it told me that turquoise for me is green. Which if I look at Turquoise the RGB color, that is green. If I look at Turquoise the mineral about half the time it is green and half the time blue.
Logically, a color, green etc., is a 'simple' notion and cannot be explained terms of anything simpler. With color we have to revert to a different description, here wavelength. But wavelength is not human perception (and we can't explain such perception in simpler terms).
(Yes in New York and Indiana, no in Massachusetts, and the law is silent elsewhere. Personally I believe that because the torta exists, the burrito may have some characteristics of a sandwich but should be considered a wrap)
I'd love a last step in the test where you're presented with the gradient, but before showing the distribution and the user's score. Allow the user to select where they consider their threshold, then display the final results.
A sorting interface would be another neat step! And yeah, I think most would gravitate toward the middle. Seeing how "far off" you are would be fun :)
Ooh maybe have the user slide a gradient left and right inside a window, aligning the center of the window with where they think the line is between blue and green (i.e., instruct the user to fill the window with equal amounts of green and blue).
It tells me to rotate my device, implying it should work on my phone, but I can't figure out how to move the colors. Holding and sliding doesn't work. Tapping doesn't seem to do anything.
> The most interesting thing for me is that while cyan (#00ffff) is nominally halfway between blue and green, most people's thresholds, averaged over monitor calibrations, imply that cyan is classified as blue.
Perceptually (that is, in CIE-LCh color space, for example), the hue component of #00ffff is a lot cloer to #00ff00 than it is to #0000ff. But the website doesn't ask which color is closer, it asks if it's "green" or "blue". And how we use those words has more to do with culture than with perception. We also call the color of a clear afternoon sky "blue", even though that is perceptually extremely far away from #0000ff.
> while cyan (#00ffff) is nominally halfway between blue and green, most people's thresholds, averaged over monitor calibrations, imply that cyan is classified as blue
Yes, because (at least for me) the thought went "well that's cyan, it's not really blue but if forced to pick, cyan is more like blue so I'll click that". It's like rounding up at 0.5.
>For me, if forced to pick between two choices that were not correct, I'd just pick one randomly. I think this is a wording problem more than anything.
That's what I'd do if I were being paid to take the survey. Instead I just closed the window as soon as it popped up cyan and only gave me blue and green as options.
But, before digital color displays became popular, the average person had, by far, mostly exposure to subtractive (paint) colors.
US school children are taught from birth that the primary subtractive colors are red, yellow, and blue, simply because those words are easier to pronounce, and so magenta is a weird "red" and cyan is a weird "blue" , until the children discover on their own, or in specialized print/paint schools, red and blue are not primary subtractive colors.
Humans are terrible at naming things.
And to bring it back to Current Thing:
Google AI cites this source for its red/yellow/blue claim, even though explicitly this source says that Google gives the wrong answer.
I mean, I was taught in grade school that George Washington cut down a cherry tree and then said he couldn't tell a lie. That didn't make it true.
I would hope that here on HN, people are aware of RGB primaries, and then maybe CMYK. Saying that cyan is "not primary or secondary" is just wrong. Even Wikipedia explains in the first paragraph that the RYB model has a "lack of scientific basis":
By the way, "cyan" is a very poor name to use for #00ffff. The term "cyan" refers to the kind of slightly greenish blue used in 4-color printing (CMYK), and was just a Greek word for "blue" chosen to be a jargon word to avoid confusion with the English color name. It has a totally different color than the equal mixture of typical G and B primaries in a computer display.
Similarly, "magenta" is a poor name to use for #ff00ff. The term "magenta" is a jargon word for the slightly purplish printer's red, which was chosen to avoid confusion with the English word "red". It has a completely different than the equal mix of RGB R and B primaries.
("Red", "green", and "blue" are also very poor names for the RGB primaries, which are substantially orangish red, yellowish green, and purplish blue.)
OP have you considered doing a version for this to test contemporary Greek native speakers, vs others ("control" group),
for differentiation of blues?
I remember reading that modern Greek has two color-names for sky- and dark- blue (not sure what the prototypes are for each nor if they have hue components, maybe the "sky" blue is green-shifted?)... always been fascinated by the discussion of "weak Sapir-Whorf" around this and would be quite interested to see if there are any differences in discrimination...
The classic cognitive/perceptual psyche data to gather would be time-to-discriminate, with the prediction being that Greek speakers make faster judgement because they have higher/faster discrimination, than others.
Not sure how you'd pose the question to non-Greek speakers tho :)
I checked in at hue 174, the median, which is interesting to me as I know that my wife will test to a very different hue as we have occasional disagreements on whether something is 'blue' or 'green' :)
It is interesting to test people at just one device.
I used my phone on a mount, and completed the test with my wife, children and myself - I was interested (though not surprised) what an outlier I was, as I am colour blind in various combinations, but though my wife scored 'bang in the middle' - it was interesting that wasn't common.
My kids were both to the left of the scale fwiw - I was further right than 98% of people.
> 2. If you average over large populations, you can estimate population thresholds, marginalizing over monitor calibrations.
This might be one case where it might make sense to cluster between the reported operating system. At the moment I only have a family of Macs to test, but I can imagine that Windows users with their different default gamma get back different results.
> I was not expecting that the median threshold (hue 174) would be so deep into the greens.
You're not asking gender of the test taker. Your results will be skewed because you're probably getting more men than women. Women in general have more ability to detect green vs blue.
Even more fundamentally, red-green colorblindness is a recessive trait on the X chromosome, thereby affecting biological males in far greater number than females.
It could be a high enough percentage to make the results from this site noticeably different between the sexes.
Not that surprising. To most people, pure RGB-blue looks a bit violet. People are used to ink (subtractive) blue more than light (additive) blue. People call the sky blue and water blue; both are closer to cyan. Most people think of a neutral blue as something like #0080ff.
I classified cyan as green because, well, it's greener than pure blue, and it's also the most greener you can get than blue, in RGB space, without losing any blue :)
>most people's thresholds, averaged over monitor calibrations, imply that cyan is classified as blue.
I think that's just to your test forcing people to pick either blue or green even though cyan is both, they are just going to pick blue because it's the first option and more likely to be picked randomly.
Color vision theory is far too complicated to discuss here, and I'm not going to debate cyan as a mixed color of blue and green wavelengths versus a fixed wavelength that's in between both of them.
What the author provided was, at best, misleading but nonsense as far as science is concerned.
If the author said he was an artist and presented colors as a preferential list it would have been a different matter.
BTW, I don't mind being voted down (it happens to me regularly), but here those who did are only showing their ignorance. I'd add the author—who penned here—ought to explain his actions in much more detail.
Not to be mean, but I think every assertion in your comment is wrong.
Blue and Green are English words which sometimes describe primary or secondary colors additive colors. Cyan is (an English word that describes) a primary subtractive color.
Colors are not English words. They're physical reactions inside our eye-brain systems, affected by varying wavelengths of light. (Actually that's not the most accurate description of color either, but it's a more useful model.)
FAQ:
* I can't know your monitor's calibration, your ambient light, or your phone's brightness. Obviously, this will affect the results. However, I am tracking local time of day and device type, from which we should be able to infer whether night mode and default calibration has any aggregate effects. Anecdotally, thus far, I haven't found any effects of Android vs. iPhone (N=34,000).
* The order is randomized. Where you start from can influence the outcome, but methodologically it's better to randomize so the aggregate results average over starting point. You can run the test several times to see how reliable this is for you.
* It's common practice in psychophysics to use two alternatives rather than three (e.g. blue, green, something in the middle). It would be a fun extension, which you can handle with an ordered logistic regression. The code is open if you want to take a shot at it: https://github.com/patrickmineault/ismyblue
* I will release aggregate results on my blog, https://neuroai.science
* I am aware of most of the limitations of this test. I have run psychophysics experiments in a lab on calibrated CRTs during my PhD in visual neuroscience. *This is just entertainment*. I did this project to see if I could make a fun webapp in Vue.js using Claude Sonnet, and later cursor, given that I am not highly proficient in modern webdev. A secondary point was to engage people in vision science and get them to talk and think about perception and language. I think it worked!