#0000FF

August 31, 2021 1:08 PM

Why are hyperlinks blue? "The internet has ingrained itself into every aspect of our lives, but there’s one aspect of the digital world that I bet you take for granted. Did you ever notice that many links, specifically hyperlinks, are blue?" Elise Blanchard investigates in an article on the Mozilla blog.
posted by jazon (43 comments total)

"What happened in 1993 to suddenly make hyperlinks blue? No one knows, but I have some theories."

Some interesting history, but there is no definitive "THIS is why!"
posted by davidmsc at 1:25 PM on August 31, 2021 [1 favorite]

Why are hyperlinks blue?

Because their baby left them?
posted by chavenet at 1:32 PM on August 31, 2021 [33 favorites]

"THIS is why"
posted by elkevelvet at 1:34 PM on August 31, 2021

Amazing article.

Aside: As someone who has done years of user experience testing, including with people with cognitive disabilites, I wish designers would either keep links blue or keep them underlined - preferrably both. I've seen so many people struggle to recognise links that were "designed" without these two basic signifiers in place.
posted by greenhornet at 1:43 PM on August 31, 2021 [28 favorites]

Blue just makes sense, because it's perceptually closer to black. So if you didn't have a color monitor (or had a color vision deficiency) there wouldn't be a big difference if you kept the underline or whatever other style changes.

But few graphics cards had more than 16 colors, so you just had to pick two that contrasted. For example, Turbo Pascal 5.0's online help had yellow hyperlinks on a dark cyan background.
posted by RobotVoodooPower at 1:46 PM on August 31, 2021 [2 favorites]

Well, I guess I'm old, because I'm reading articles guessing about history I lived through:
I think the real reason why we have blue hyperlinks is simply because color monitors were becoming more popular around this time.

It's like saying TikTok is so popular in 2020 because smartphones were becoming more popular around this time.

The blue hyperlink is from 1993; the first IBM PC clones had colour graphics adapters (CGA) which sucked but did show CMYK (not the CMYK model, just those four colours -- 16 colours in text mode) in 1981. By 1987, VGA which handled 256 colours, was the new standard and it was ubiquitous by 1990. If you came of computer age post 2000, you can't appreciate how quickly the technology was advancing and how frequently systems were upgraded. This link shows the start of King's Quest in CGA, EGA (the intermediate 1984 standard) and VGA; it's night and day.

In fact, the monochrome Mac was peaking as a percent of sales (only at 10%) -- 1993 was actually a local minimum of colour computers.
posted by Superilla at 2:19 PM on August 31, 2021 [25 favorites]

My first real world introduction to clever hacks was on the Hotwired website in something like 1994. The homepage was vibrant and they had somehow figured out how to remove the blue borders around the image links! It took me a long time to realise that they had made the background colour of the page the same colour blue as hyperlinks. I thought it was genius.
posted by Foaf at 2:24 PM on August 31, 2021 [4 favorites]

Roses are #FF0000
Violets are #0000FF
All my links are belong to you!

(sorry...)
posted by Insert Clever Name Here at 2:24 PM on August 31, 2021 [14 favorites]

I wonder if the phrase "That link stays blue" will survive into the future just like "I dialed a phone number"?
posted by loquacious at 2:27 PM on August 31, 2021 [2 favorites]

I've seen so many people struggle to recognize links that were "designed" without these two basic signifiers in place.

Not just people with obvious cognitive disabilities or novice computer users. I'm not an expert on web stuff, but I'm starting out ahead of many people. (I've hacked a few browser extensions, maintain several websites and wikies, have fancy STEM credentials, etc.)

Last year I spent around 15 minutes on the phone with my credit union because I couldn't figure out how to log into their website. I went around and around with their tech support in utter confusion. I tried four browsers on two OSes, checked that it wasn't a spoofed site, etc. It turns out that I just didn't even consider that the colored box around the word "account" was a link. I just assumed it was a decorated label for the drop-down menu next to it, which was actually the place to make new accounts of various kinds. I'd see the word account, use the menu to select the appropriate kind of account, put in my username, and get told that I already had an account. I now have notes in my password manager about how to log in, since I do it so rarely.
posted by eotvos at 2:38 PM on August 31, 2021 [8 favorites]

While the overall conclusion of "Mosaic did it and everyone copied it" is likely correct and matches my recollection, some of the other facts are not.

WWW, the first browser, was created in 1987

That's a few years too early. First release was 1990.

Gopher Protocol's ... original design featured green text on a black background.

This is in the not-even-wrong category of mistakes.
posted by autopilot at 3:25 PM on August 31, 2021 [32 favorites]

My theory is that Psychological research has shown that Blue and Red are colors that we respond to the most. As Red usually is used for caution and warning; websites do the same. Blue is more neutral but still stands out. Learned this from Thinking Fast and Slow. Because of this; all my work shirts are now shades of red or blue. :)
posted by indianbadger1 at 3:28 PM on August 31, 2021 [2 favorites]

> The blue hyperlink is from 1993; the first IBM PC clones had colour graphics adapters (CGA) which sucked but did show CMYK (not the CMYK model, just those four colours -- 16 colours in text mode) in 1981. By 1987, VGA which handled 256 colours, was the new standard and it was ubiquitous by 1990. If you came of computer age post 2000, you can't appreciate how quickly the technology was advancing and how frequently systems were upgraded.

I was designing and building websites for Fortune 500 companies starting in late 1994. On the one hand, device testing at the time was straightforward compared to now: Mac System 7 & Windows 3.1, Mosaic and Netscape on each*... On the other hand, it also meant testing at 1, 4, and 8 bit color depths. 1-bit is black and white, simple enough. 8-bit is kinda complicated but could be managed.

The nightmare was supporting 4-bit video displays. This meant 16 colors, which could take the very minimal control you already had over web page presentations and smash it repeatedly with a large hammer until the page was reduced to pixelated shrapnel of yellows, blues and greens.

*(Sorry, Unix users of the mid-90s.)
posted by ardgedee at 3:50 PM on August 31, 2021 [8 favorites]

Aside: As someone who has done years of user experience testing, including with people with cognitive disabilites, I wish designers would either keep links blue or keep them underlined - preferrably both.

I am a designer and let me tell you, I have never met another designer who understands that everything that is clickable must be colored in some way to let the user know it's clickable. Once I'm able to get that concept across then we can talk about and they should be blue. And please don't make things blue that are not clickable. I alone am fighting the good fight over here.
posted by bleep at 3:55 PM on August 31, 2021 [31 favorites]

It's like saying TikTok is so popular in 2020 because smartphones were becoming more popular around this time.

Yeah, the only non-color computer we ever had was the original TRS-80 (1977), which later did have a color model. The Apple II (1977) and C64 (1982) were color computers, for example.
posted by thefoxgod at 4:14 PM on August 31, 2021

But bleep, everything on a website should be light gray text on a white background, because it looks SO COOL on a perfectly-balanced monitor under ideal lighting conditions! Why would I want things like hover or focus states? They're ruining The Vision!
posted by fifteen schnitzengruben is my limit at 4:57 PM on August 31, 2021 [5 favorites]

> I wonder if the phrase "That link stays blue" will survive into the future just like "I dialed a phone number"?

I've been around since USENET (and heck, x/y/zmodem) and I've never heard that phrase before.
posted by stevil at 5:05 PM on August 31, 2021 [15 favorites]

On the other hand, it also meant testing at 1, 4, and 8 bit color depths

Color dithering used to be a thing on the web. Not needed any longer, but all the algorithms behind it are interesting stuff. Here's a fun overview of how dithering can be used for making 1-bit B+W images that simulate greyscale.
posted by They sucked his brains out! at 5:08 PM on August 31, 2021 [1 favorite]

Color dithering was often (when not cleverly designed for) surprisingly useless! Keeping in mind this was an era of 72 pixel per inch displays for Macs. The Windows standard was nominally better -- 96 ppi, iirc -- but in practice on any display dots were large enough to be readable as discrete dots rather than pointillistic elements. The color shifts you get at the edges of antialiased curves and diagonals really stand out when the dots are the size of chocolate jimmies.

(And yes in reality no monitor was exactly 72 or 96 ppi, but the displays were all smaller and the dots a lot larger than common displays today. You can't appreciate how things look at 640 x 480 pixels until you try.)

The other problem with dithering was that mesh patterns made very inefficient GIFs -- the LZW compression algorithm optimally handles adjacent pixels with the same color, so an image with clever dithering could be three times larger than one that didn't look as good but mostly consisted of large flat areas of color -- when a page banner might already take five or ten seconds to load over a 14400 baud modem, loading three times slower is brutal. And manually dithering JPGs was pointless since the compression algorithm was going to average out all the colors.
posted by ardgedee at 5:56 PM on August 31, 2021 [2 favorites]

None of the top 10 websites consistently use blue to denote a link. Now it is sort of an Adverture Game style clickfest to see if something is clickable. For some reason my mind knows that text that is a different size or shape is going to likely be a link but beyond there there's no common signifiers. Even Google is not consistent.
posted by geoff. at 6:26 PM on August 31, 2021 [2 favorites]

The worst are sites that don't differentiate links at all except for those linking to advertising which they make standard blue even when you've already clicked them.
posted by Mitheral at 6:33 PM on August 31, 2021 [1 favorite]

"Why are hyperlinks blue?" the hyperlink asked greenly ... at least in my preferences ... embodying irony itself!
posted by traveler_ at 7:31 PM on August 31, 2021 [1 favorite]

Anybody know what the UX for Lotus Notes looked like around that time?
I was trying to think of other software from the period which might have similar design elements, and that came to mind.
posted by cheshyre at 7:48 PM on August 31, 2021 [1 favorite]

The Apple II (1977) and C64 (1982) were color computers, for example.

Even the TI-99/4A was colour!

I have no point here other than to reflexively note the existence of the TI-99/4A whenever the Apple ][ and C64/Vic20 are mentioned in casual conversation. Old Fidowars surprise and embarrass you at the strangest times.
posted by bonehead at 7:58 PM on August 31, 2021 [4 favorites]

And I simply must share this rabbithole I just discovered:

Ben Kraal publishes a newsletter about User Experience as seen through the lens of academic papers from 1992 - right around the time we're looking at.

Newsletter, and a recent episode of the UX Podcast talking with him
posted by cheshyre at 7:59 PM on August 31, 2021 [1 favorite]

Color computers were common in the early 90s but they weren't evenly distributed. In particular, the high-end workstations used for enterprise software development lagged a bit.

Looking at this photo of Tim Berners-Lee's original NeXT Computer, on which he wrote "WorldWideWeb", I am pretty sure it's a monochrome system.

The first-generation NeXT boxes, which were just called the "NeXT Computer", were monochrome-only. NeXT didn't support color until their second generation of machines, which weren't shipped until September of 1990. (And even then, color was an optional add-on to both the NeXT Cube and the NeXTstation.)

Unfortunately, I can't find any screenshots of the initial version of TBL's WorldWideWeb; most articles about it use screenshots of later versions (e.g. this one from 1994) running on much later NeXT hardware or modern emulators. But I'm pretty sure that in its original version hyperlinks were just underlined.

ViolaWWW, which was released in early 1992 and was the first popular web browser for Unix/Xwindows, was also monochrome in its first incarnation, featuring black text on a grey background, with only underlining to indicate links. This is consistent with much of the Unix workstation hardware of the time (like the VAXstations), which often had color graphics and monitors as high-priced extra options.

It makes sense that Mosaic was probably responsible for adding the default blue color to links, since they were the first browser to display inline graphics. Which suggests to me that they were probably developing on, and intended the software be used on, color machines. And that would be consistent with the timeline: by early 1993, prices of color monitors had come down enough that they were becoming standard equipment everywhere—even Apple went to color in its compact Macs with the Color Classic in February of that year.

My strong feeling is that blue was chosen for links because it's a color that's nearly guaranteed to be on any system (even IBM CGA machines have light and dark blue) and is relatively easy to read on any background that black text would be—something that you can't say about a lot of the other CGA / 16-color-palette color choices.
posted by Kadin2048 at 8:17 PM on August 31, 2021 [1 favorite]

More than 25 years ago a colleague made links orange in a website we were making. I theft them a note to “make all the links 0000FF!, no questions no arguments it’s the law”.

On Monday link text had been changed to “ooooff!”
posted by Dr. Curare at 8:18 PM on August 31, 2021 [14 favorites]

Dr. Curare, that reminds me of the racing horse called Potoooooooo.
posted by sixswitch at 9:41 PM on August 31, 2021 [2 favorites]

Let's talk about sex:

Sex differences in response to red and blue light in human primary visual cortex: a bold fMRI study
[I can only hope making BOLD lower case in the title was a typo]

As there are numerous sex-related differences in central nervous system dopamine function, we predicted that blue and red light stimulation would produce sex-specific patterns of response in primary visual cortex when studied using the blood oxygen level dependent (BOLD) functional magnetic resonance imaging (fMRI) technique. We analyzed the BOLD response to red and blue light in male and female human volunteers (N=20). Red and blue light responses in primary visual cortex (V1) to stepped intensities of red and blue light were compared by sex for threshold to detectable BOLD signal increase and for stimulus intensity vs. BOLD signal response. Near threshold, males and females showed similar BOLD signal change to red light, but males showed a threefold greater increase (0.52%) to blue light stimulation when compared to females (0.14%). Log-linear regression modeling revealed that the slope coefficients for the red light stimulus intensity vs. signal change curve were not significantly different for males and females (z=0.995, P=0.320), whereas the slope coefficients for the blue light stimulus intensity vs. signal change curve were significantly larger in males (z=2.251, P=0.024). These findings support a sex and color-dependent differential pattern of primary visual cortical response to photic stimulation and suggest a method for assessing the influence of specific dopamine agonist/antagonist medications on visual function.

So males are much more sensitive to blue light at "threshold" than females are, and making links blue is just one more way in which the internet was a boyzone back then.

But I think it goes further than that. We all know that Viagra enhances the perception of blue; so much so that the FAA forbids pilots from taking it a certain number of hours before a flight because it distorts their perception of colors to a degree which is considered dangerous. When boys reach puberty and start having involuntary erections, it happens because their endogenous systems with a Viagra-like effect become more active, and I can't see how that could fail to make blue pop out at those boys more than it did before.

And considering the demographic which is popularly associated with making the internet a thing, I'm forced to conclude that links are blue because blue is the color of erections.

Which means that you are pretty much right on target, chavenet:

Why are hyperlinks blue?

Because their baby left them?

Either that, or they're still looking around for that first one.
posted by jamjam at 10:31 PM on August 31, 2021 [5 favorites]

This is the kind of elaborate, ungrounded but entertaining speculation based on esoteric kernels of truth that I come to Metafilter for. I salute you, jamjam.
posted by i_am_joe's_spleen at 10:56 PM on August 31, 2021 [5 favorites]

indianbadger1: My theory is that Psychological research has shown that Blue and Red are colors that we respond to the most. As Red usually is used for caution and warning; websites do the same. Blue is more neutral but still stands out.

This makes total sense. And it's not like there were a lot of other options. Red means stop, green means go; blue was available, and neutral but noticeable. It's good if links are easy to spot.

I really think it's as simple as that.
posted by Too-Ticky at 3:14 AM on September 1, 2021 [2 favorites]

I'm curious about non-browser links, which she doesn't discuss.
Decades ago I used Bookmaster, and the links there were blue (and underlined), although I can't verify that it was before 1993.
I can't find out too much about its history, although it came out in 1980, and by then there were color mainframe terminals.
posted by MtDewd at 4:41 AM on September 1, 2021 [1 favorite]

Didn't the help system in Windows 3.1 have blue hyperlinks? Or were they green?
posted by RonButNotStupid at 4:48 AM on September 1, 2021

Blue just makes sense, because it's perceptually closer to black.

Having gone down a few of these ratholes myself, you really need to be wary of arguments like this. There's always something a lot more useful behind them than a just-so story.
posted by mhoye at 5:38 AM on September 1, 2021 [1 favorite]

>the monochrome Mac was peaking as a percent of sales (only at 10%) -- 1993

?? The only non-color Macs being sold in '93 were the Power books

https://everymac.com/systems/by_year/macs-released-in-1993.html

I've never seen Windows running on a monochrome monitor for that matter.
posted by Heywood Mogroot III at 7:16 AM on September 1, 2021

> Didn't the help system in Windows 3.1 have blue hyperlinks? Or were they green?

Links were the same color as text but underlined, therefore black by default and changeable depending on your choice of color theme. You can see for yourself in an emulator because computers are so much zoomier now.
posted by ardgedee at 8:41 AM on September 1, 2021

The hacker news discussion really is good. Likely answer afaict: on a CRT it was going to be one of RGBK*, and Ben Schneiderman did actual experimental user testing before picking blue**. Red was more findable but reduced reading comprehension of the text itself.

*anything else makes blurrier text on a misaligned CRT, which were common

**Bit I like about this - they weren’t just figuring out what color to use for inline text links, they were figuring out if inline text links would work at all. Needs a pointer and mice were still new.
posted by clew at 9:09 AM on September 1, 2021 [1 favorite]

Parenthetical to be more scrupulous: inline links don’t need a pointing device. Without a pointer, the advantage of inline links over a keyboard-accessible menu mode designed for keyboard navigation is a lot less obvious, especially in 1988.
posted by clew at 9:28 AM on September 1, 2021

The only non-color Macs being sold in '93 were the Power books

Those are new models released in 1993. The actual models on sale that year would have been 1992 models which still included monochrome screens. Desktops could still be bought with monochrome screens and there would be many many older monochrome Macs still in use.
posted by grahamparks at 9:35 AM on September 1, 2021 [1 favorite]

I've been around since USENET (and heck, x/y/zmodem) and I've never heard that phrase before.

It's relatively recent, say the last 5-10 years.

The phrase is used for risky or questionable links or being rickrolled, or, ugh, goatse'ed or something and it basically means "Yeah I'm not clicking on your link."

It's really common on reddit and other forums that are, ah, less moderated.
posted by loquacious at 9:39 AM on September 1, 2021

I think it was around 2011 that mobile designers gave up on affordances and consistency in UX. Maybe it was the rise of Snapchat, but it also seems to coincide with the passing of Steve Jobs, so ... (gestures broadly at everything, including the "shake to undo" gesture, activating many hidden app features)
posted by RobotVoodooPower at 11:55 AM on September 1, 2021 [1 favorite]

Oh boy, jamjam, that paper is entertaining reading. It's like a greatest hits of everything that was bad about fMRI studies in the late 1990's and early 2000's.

My favorite bit so far is this, which just might take the lead in my personal ranking for Most Unnecessary and Inappropriate Use of a P-Value (a ranking I have just decided to start keeping): "The mean age for males was significantly higher than for females in both cases, P<0.001." Note that there were only 8 and 11 subjects in the two cases they're discussing. I was going to complain that there's no mention made of multiple comparisons correction, which was uncommon to do at the time and is a good reason that marginal results like that p=.024 they're hanging their study on should be viewed with extreme side-eye. But if they're running significance tests on the age of the subjects in groups that they themselves assigned, complaining about a lack of multiple comparisons correction seems almost unfair. Probably they calculated that p-value just to get one of the reviewers to stop complaining about their pet theory that it was all age-related or something, but still.
posted by biogeo at 4:06 PM on September 1, 2021 [2 favorites]

Didn't the help system in Windows 3.1 have blue hyperlinks? Or were they green?

Green links were for tooltippy definitional popups. Windows 95 and beyond had it, but I'm not sure Windows 3.1 had more than a single link type.
posted by rhizome at 4:24 PM on September 1, 2021