Eric Meyer gift wraps the most awkwardly shaped of boxes using nothing but CSS, HTML and a little curl of ribbon. No matter how well you plan and how much paper you have at your disposal, sometimes you just need to slant the gift to the side.

We have a lot of new layout tools at our disposal these days—flexbox is finally stable and interoperable, and Grid very much the same, with both technologies having well over 90% support coverage. In that light, we might think there’s no place for old tricks like negative margins, but I recently discovered otherwise.

Over at An Event Apart, we’ve been updating some of our landing pages, and our designer thought it would be interesting to have slanted images of speakers at the tops of pages. The end result looks like this.

The interesting part is the images. I wanted to set up a structure like the following, so that it will be easy to change speakers from time to time while preserving accessible content structures:

<div id="page-top">
  <ul class="monoliths">
    <li>
      <a href="https://aneventapart.com/speakers/rachel-andrew"> 
        <img src="/img/rachel-andrew.jpg" alt=""> 
        <div> 
          <strong>Rachel Andrew</strong> CSS Grid 
        </div> 
      </a>
    </li>
    <li>
      <a href="https://aneventapart.com/speakers/derek-featherstone"> 
        <img src="/img/derek-featherstone.jpg" alt=""> 
        <div> 
          <strong>Derek Featherstone</strong> Accessibility 
        </div> 
      </a>
    </li>
    <li>
      …
    </li>
    <li>
      …
    </li>
  </ul>
</div>

The id value for the div is straightforward enough, and I called the ul element monoliths because it reminded me of the memorial monoliths at the entrance to EPCOT in Florida. I’m also taking advantage of the now-ubiquitous ability to wrap multiple elements, including block elements, in a hyperlink. That way I can shove the image and text structures in there, and make the entire image and text below it one link.

Structure is easy, though. Can we make that layout fully responsive? I wondered. Yes we can. Here’s the target layout, stripped of the navbar and promo copy.

So let’s start from the beginning. The div gets some color and text styling, and the monoliths list is set to flex. The images are in a single line, after all, and I want them to be flexible for responsive reasons, so flexbox is 100% the right tool for this particular job.

#page-top { 
  background: #000; 
  color: #FFF; 
  line-height: 1; 
} 
#page-top .monoliths { 
  display: flex; 
  padding-bottom: 1em; 
  overflow: hidden; 
}

I also figured, let’s give the images a simple basis for sizing, and set up the hyperlink while we’re at it.

#page-top .monoliths li { 
  width: 25%; 
} 
#page-top .monoliths a { 
  color: inherit; 
  text-decoration: inherit; 
  display: block; 
  padding: 1px; 
}

So now the list items are 25% wide—I can say that because I know there will be four of them—and the links pick up the foreground color from their parent element. They’re also set to generate a block box.

At this point, I could concentrate on the images. They need to be as wide as their parent element, but no wider, and also match height. While I was at it, I figured I’d create a little bit of space above and below the captioning text, and make the strong elements containing speakers’ names generate a block box.

#page-top .monoliths img { 
  display: block; 
  height: 33rem; 
  width: 100%; 
} 
#page-top .monoliths div { 
  padding: 0.5em 0; 
} 
#page-top .monoliths strong { 
  display: block; 
  font-weight: 900; 
}

It looks like the speakers were all cast into the Phantom Zone or something, so that needs to be fixed. I can’t physically crop the images to be the “correct” size, because there is no correct size: this needs to work across all screen widths. So rather than try to swap carefully-sized images in and out at various breakpoints, or complicate the structure with a wrapper element set to suppress overflow of resized images, I turned to object-fit.

#page-top .monoliths img { 
  display: block; 
  height: 33rem; 
  width: 100%; 
  object-fit: cover; 
  object-position: 50% 20%; 
}

If you’ve never used object-fit, it’s a bit like background-size. You can use it to resize image content within the image’s element box without creating distortions. Here, I set the fit sizing to cover, which means all of the img element’s element box will be covered by image content. In this case, it’s like zooming in on the image content. I also set a zooming origin with object-position, figuring that 50% across and 20% down would be in the vicinity of a speaker’s face, given the way pictures of people are usually taken.

This is fairly presentable as-is—a little basic, perhaps, but it would be fine to layer the navbar and promo copy back over it with Grid or whatever, and call it a day. But it’s too square and boxy. We must go further!

To make that happen, I’m going to take out the third and fourth images temporarily, so we can see more clearly how the next part works. That will leave us with Rachel and Derek.

The idea here is to clip the images to be slanted, and then pull them close to each other so they have just a little space between them. The first part is managed with clip-path, but we don’t want to pull the images together unless their shapes are being clipped. So we set up a feature query.

@supports (clip-path: polygon(0 0)) or (-webkit-clip-path: polygon(0 0)) { 
  #page-top .monoliths li { 
    width: 37.5%; 
  } 
}

I decided to test for both the un-prefixed and WebKit-prefixed versions of clip-path because Safari still requires the prefix, and I couldn’t think of a good reason to penalize Safari’s users for the slowness of its standards advancement. Then I made the images wider, taking them from 25% to 37.5%, which makes them half again as wide.

Thanks to object fitting, the images don’t distort when I change their parent’s width; they just get wider and scale up the contents to fit. And now, it is time for clipping!

@supports (clip-path: polygon(0 0)) or (-webkit-clip-path: polygon(0 0)) { 
  #page-top .monoliths li { 
    width: 37.5%; 
    -webkit-clip-path: polygon(25% 0, 100% 0, 75% 100%, 0 100%); 
    clip-path: polygon(25% 0, 100% 0, 75% 100%, 0 100%); 
  } 
}

Each coordinate pair in the polygon() is like the position pairs in background-position or object-position: the horizontal distance first, followed by the vertical distance. So the first point in the polygon is 25% 0, which is 25% of the way across the element box, and no distance down, so right at the top edge. 100% 0 is the top right corner. 75% 100% is on the bottom edge, three-quarters of the way across the element, and 0 100% is the bottom left corner. That creates a polygon that’s a strip three-quarters the full width of the element box, and runs from bottom left to top right.

Now we just have to pull them together, and this is where old tricks come back into play: all we need is a negative right margin to bring them closer together.

#page-top .monoliths li { 
  width: 37.5%; 
  margin-right: -7.5%; 
  -webkit-clip-path: polygon(25% 0, 100% 0, 75% 100%, 0 100%); 
  clip-path: polygon(25% 0, 100% 0, 75% 100%, 0 100%); 
}

The separation between them is a little wider than we were originally aiming for, but let’s see what happens when we add the other two images back in and let flexbox do its resizing magic.

Notice how the slants actually change shape as the screen gets narrower or wider. This is because they’re still three-quarters the width of the image element’s box, but the width of that box is changing as the screen width changes. That means at narrow widths, the slant is much steeper, whereas at wide widths, the slant is more shallow. But since the clipping path’s coordinates were all set with percentage distances, they all stay parallel to each other while being completely responsive to changes in screen size. An absolute measure like pixels would have failed.

But how did the images get closer together just by adding in two more? Because the list items’ basic sizing added up to more than 100%, and they’re all set to flex-shrink: 1. No, you didn’t miss a line in the CSS: 1 is the default value for flex-shrink. Flex items will shrink by default, which after all is what we should expect from a flexible element. If you want to know how much they shrunk, and why, here’s what Firefox’s flex inspector reports.

When there were only two list items, there was space enough for both to be at their base size, with no shrinkage. Once we went to four list items, there wasn’t enough space, so they all shrank down. At that point, having a negative right margin of -7.5% was just right to pull them together to act as a unit.

So, now they’re all nicely nestled together, and fully responsive! The captions need a little work, though. Notice how they’re clipped off a bit on the left edge, and can be very much clipped off on the right side at narrower screen widths? This happens because the li elements are being clipped, and that clipping applies to all their contents, images and text alike. And we can’t use overflow to alter this: clipped is clipped, not overflowed.

Fortunately, all we really need to do is push the text over a small amount. Inside the feature query, I added:

#page-top .monoliths div { 
  padding-left: 2%;
  padding-right: 26%; 
}

This shifts the text just a bit rightward, enough to clear the clip path. On the right side, I padded the div boxes so their contents wouldn’t fall outside the clipped area and appear to slide under the next caption. We could also use margins here, but I didn’t for reasons I’ll make clear at the end.

At the last minute, I decided to make the text at least appear to follow the slants of the images. For that, I just needed to shift the first line over a bit, which I did with a bit more padding.

#page-top .monoliths strong { 
  padding-left: 1%; 
}

That’s all to the good, but you may have noticed the captions still overlap at really narrow screen widths. There are a lot of options here, from stacking the images atop one another to reverting to normal flow, but I decided to just hide the captions if things got too narrow. It reduces clutter without sacrificing too much in the way of content, and by leaving them still technically visible, they seem to remain accessible.

@media (max-width: 35rem) { 
  #page-top .monoliths div { 
    opacity: 0.01 
  } 
}

And that, as they say, is that! Fully responsive slanted images with text, in an accessible markup structure. I dig it.

I did fiddle around with the separations a bit, and found that a nice thin separator occurred around margin-right: -8%, whereas beefier ones could be found above -7%. And if you crank the negative margin value to something beyond -8%, you’ll make the images overlap entirely, no visible separation—which can be a useful effect in its own right.

I promised to say why I used padding for the caption text div rather than margins. Here’s why.

#page-top .monoliths div { 
  padding-left: 3%; 
  padding-right: 26%; 
  border-top: 2px solid transparent; 
  background: linear-gradient(100deg,hsl(292deg,50%,50%) 50%, transparent 85%); 
  background-clip: padding-box; 
}

It required a wee bit more padding on the left to look decent, and an alteration to the background clipping box in order to keep the purple from filling the transparent border area, but the end result is pretty nifty, if I do say so myself. Alternatively, we could drop the background gradient on the captions and put one in the background, with a result like this.

I have no doubt this technique could be extended, made more powerful, and generally improved upon. I really wished for subgrid support in Chrome, so that I could put everything on a grid without having to tear the markup structure apart, and there are doubtless even more interesting clipping paths and layout patterns to try out.

I hope these few ideas spark some much better ideas in you, and that you’ll share them with us!

About the author

Eric A. Meyer (@meyerweb) has been a burger flipper, a college webmaster, an early blogger, one of the original CSS Samurai, a member of the CSS Working Group, a consultant and trainer, and a Standards Evangelist for Netscape. Among other things, Eric co-wrote Design For Real Life with Sara Wachter-Boettcher for A Book Apart and CSS: The Definitive Guide with Estelle Weyl for O’Reilly, created the first official W3C test suite, assisted in the creation of microformats, and co-founded An Event Apart with Jeffrey Zeldman. Eric lives with his family in Cleveland, Ohio, which is a much nicer city than you’ve probably heard. He enjoys a good meal whenever he can and considers almost every form of music to be worthwhile.

More articles by Eric

Divya Sasidharan calls into question the trade-offs often made between security and usability. Does a secure interface by necessity need to be hard to use? Or is it the choice we make based on years of habit? Snow has fallen, snow on snow.

Security is often synonymous with poor usability. We assume that in order for something to be secure, it needs to by default appear impenetrable to disincentivize potential bad actors. While this premise is true in many instances like in the security of a bank, it relies on a fundamental assumption: that there is no room for choice.

With the option to choose, a user almost inevitably picks a more usable system or adapts how they interact with it regardless of how insecure it may be. In the context of the web, passwords are a prime example of such behavior. Though passwords were implemented as a way to drastically reduce the risk of attack, they proved to be marginally effective. In the name of convenience, complex, more secure passwords were shirked in favor of easy to remember ones, and passwords were liberally reused across accounts. This example clearly illustrates that usability and security are not mutually exclusive. Rather, security depends on usability, and it is imperative to get user buy-in in order to properly secure our applications.

Security and Usability; a tale of broken trust

At its core, security is about fostering trust. In addition to protecting user accounts from malicious attacks, security protocols provide users with the peace of mind that their accounts and personal information is safe. Ironically, that peace of mind is incumbent on users using the security protocols in the first place, which further relies on them accepting that security is needed. With the increased frequency of cyber security threats and data breaches over the last couple of years, users have grown to be less trusting of security experts and their measures. Security experts have equally become less trusting of users, and see them as the “the weakest link in the chain”. This has led to more cumbersome security practices such as mandatory 2FA and constant re-login flows which bottlenecks users from accomplishing essential tasks. Because of this break down in trust, there is a natural inclination to shortcut security altogether.

Build a culture of trust not fear

Building trust among users requires empowering them to believe that their individual actions have a larger impact on the security of the overall organization. If a user understands that their behavior can put critical resources of an organization at risk, they will more likely behave with security in mind. For this to work, nuance is key. Deeming that every resource needs a similarly high number of checks and balances diminishes how users perceive security and adds unnecessary bottlenecks to user workflows.

In order to lay the foundation for good security, it’s worth noting that risk analysis is the bedrock of security design. Instead of blindly implementing standard security measures recommended by the experts, a better approach is to tailor security protocols to meet specific use cases and adapt as much as possible to user workflows. Here are some examples of how to do just that:

Risk based authentication

Risk based authentication is a powerful way to perform a holistic assessment of the threats facing an organization. Risks occur at the intersection of vulnerability and threat. A high risk account is vulnerable and faces the very real threat of a potential breach. Generally, risk based authentication is about calculating a risk score associated with accounts and determining the proper approach to securing it. It takes into account a combination of the likelihood that that risk will materialize and the impact on the organization should the risk come to pass. With this system, an organization can easily adapt access to resources depending on how critical they are to the business; for instance, internal documentation may not warrant 2FA, while accessing business and financial records may.

Dynamically adaptive auth

Similar to risk based auth, dynamically adaptive auth adjusts to the current situation. Security can be strengthened and slackened as warranted, depending on how risky the access point is. A user accessing an account from a trusted device in a known location may be deemed low risk and therefore not in need of extra security layers. Likewise, a user exhibiting predictive patterns of use should be granted quick and easy access to resources. The ability to adapt authentication based on the most recent security profile of a user significantly improves the experience by reducing unnecessary friction.

Conclusion

Historically, security failed to take the user experience into account, putting the onus of securing accounts solely on users. Considering the fate of password security, we can neither rely on users nor stringent security mechanisms to keep our accounts safe. Instead, we should aim for security measures that give users the freedom to bypass them as needed while still protecting our accounts from attack. The fate of secure systems lies in the understanding that security is a process that must constantly adapt to face the shifting landscape of user behavior and potential threats.

About the author

Divya is a web developer who is passionate about open source and the web. She is currently a developer experience engineer at Netlify, and believes that there is a better workflow for building and deploying sites that doesn’t require a server—ask her about the JAMstack. You will most likely find her in the sunniest spot in the room with a cup of tea in hand.

More articles by Divya

Amy Hupe prepares a four bird roast of tasty treats so we can learn how the needs of many different types of users can be served through careful implementation of components within a design system.

Design systems help us to make our products consistent, and to make sure we’re creating them in the most efficient way possible. They also help us to ensure our products are designed and built to a high quality; that they’re not only consistent in appearance, and efficiently-built, but that they are good. And good design means accessible design.

1 in 5 people in the UK have a long term illness, impairment or disability – and many more have a temporary disability. Designing accessible services is incredibly important from an ethical, reputational and commercial standpoint. For EU government websites and apps, accessibility is also a legal requirement.

With that in mind, I’ll explain the four main ways I think we can use design systems to promote accessible design within an organisation, and what design systems can’t do.

1. Bake it in

Design systems typically provide guidance and examples to aid the design process, showing what best practice looks like. Many design systems also encompass code that teams can use to take these elements into production. This gives us an opportunity to build good design into the foundations of our products, not just in terms of how they look, but also how they work. For everyone.

Let me give an example.

The GOV.UK Design System contains a component called the Summary list. It’s used in a few different contexts on GOV.UK, to summarise information. It’s often used at the end of a long or complex form, to let users check their answers before they send them, like this:

Users can review the information and, if they’ve entered something incorrectly, they can go back and edit their answer by clicking the “Change” link on the right-hand side. This works well if you can see the change link, because you can see which information it corresponds to.

In the top row, for example, I can see that the link is giving me the option to change the name I’ve entered because I can see the name label, and the name I put in is next to it.

However, if you’re using a screen reader, this link – and all the others – will just say “change”, and it becomes harder to tell what you’re selecting. So to help with this, the GOV.UK Design System team added some visually-hidden text to the code in the example, to make the link more descriptive.

Sighted users won’t see this text, but when a screen reader reads out the link, it’ll say “change name”. This makes the component more accessible, and helps it to satisfy a Web Content Accessibility Guidelines (WCAG 2.1) success criterion for links which says we must “provide link text that identifies the purpose of the link without needing additional context”.

By building our components with inclusion in mind, we can make it easier to make products accessible, before anyone’s even had to think about it. And that’s a great starting point. But that doesn’t mean we don’t have to think about it – we definitely do. And a design system can help with that too.

2. Explain it

Having worked as the GOV.UK Design System’s content designer for the best part of 3 years, I’m somewhat biased about this, but I think that the most valuable aspect of a design system is its documentation.

(Here’s a shameless plug for my patterns Day talk on design system documentation earlier this year, if you want to know more about that.)

When it comes to accessibility, written documentation lets us guide good practice in a way that code and examples alone can’t.

By carefully documenting implementation rules for each component, we have an opportunity to distribute accessible design principles throughout a design system. This means design system users encounter them not just once, but repeatedly and frequently, in various contexts, which helps to build awareness over time.

For instance, WCAG 2.1 warns against using colour as “the only visual means of conveying information, calling an action, prompting a response or distinguishing a visual element”. This is a general principle to follow, but design system documentation lets us explain how this relates to specific components.

Take the GOV.UK Design System’s warning buttons. These are used for actions with serious, often destructive consequences that can’t easily be undone – like permanently deleting an account.

The example doesn’t tell you this, but the guidance explains that you shouldn’t rely on the red colour of warning buttons to communicate that the button performs a serious action, since not all users will be able to see the colour or understand what it signifies.

Instead, it says, “make sure the context and button text makes clear what will happen if the user selects it”. In this way, the colour is used as an enhancement for people who can interpret it, but it’s not necessary in order to understand it.

Making the code in our examples and component packages as accessible as possible by default is really important, but written documentation like this lets us be much more explicit about how to design accessible services.

3. Lead by example

In our design systems’ documentation, we’re telling people what good design looks like, so it’s really important that we practice what we preach.

Design systems are usually for members of staff, rather than members of the public. But if we want to build an inclusive workplace, we need to hold them to the same standards and ensure they’re accessible to everyone who might need to use them – today and in the future.

One of the ways we did this in my team, was by making sure the GOV.UK Design System supports users who need to customise the colours they use to browse the web. There are a range of different user needs for changing colours on the web. People who are sensitive to light, for instance, might find a white background too bright. And some users with dyslexia find certain colours easier to read than others.

My colleague, Nick Colley, wrote about the work we did to ensure GOV.UK Design System’s components will work when users change colours on GOV.UK. To ensure we weren’t introducing barriers to our colleagues, we also made it possible to customise colours in the GOV.UK Design System website itself.

Building this flexibility into our design system helps to support our colleagues who need it, but it also shows others that we’re committed to inclusion and removing barriers.

4. Teach it

The examples I’ve drawn on here have mostly focused on design system documentation and tooling, but design systems are much bigger than that. In the fortuitously-timed “There is No Design System”, Jina reminds us that tooling is just one of the ways we systematise design:

…it’s a lot of people-focused work: Reviewing. Advising. Organizing. Coordinating. Triaging. Educating. Supporting.”

To make a design system successful, we can’t just build a set of components and hope they work. We have to actively help people find it, use it and contribute to it. That means we have to go out and talk about it. We have to support people in learning to use it and help new teams adopt it. These engagement activities and collaborative processes that sit around it can help to promote awareness of the why, not just the what.

At GDS, we ran workshops on accessibility in the design system, getting people to browse various web pages using visual impairment simulation glasses to understand how visually impaired users might experience our content. By working closely with our systems’ users and contributors like this, we have an opportunity to bring them along on the journey of making something accessible.

We can help them to test out their code and content and understand how they’ll work on different platforms, and how they might need to be adjusted to make sure they’re accessible. We can teach them what accessibility means in practice.

These kinds of activities are invaluable in helping to promote accessible design thinking. And these kinds of lessons – when taught well – are disseminated as colleagues share knowledge with their teams, departments and the wider industry.

What design systems can’t do

Our industry’s excitement about design systems shows no signs of abating, and I’m excited about the opportunities it affords us to make accessible design the default, not an edge case. But I want to finish on a word about their limitations.

While a design system can help to promote awareness of the need to be accessible, and how to design products and services that are, a design system can’t make an organisation fundamentally care about accessibility.

Even with the help of a thoughtfully created design system, it’s still possible to make really inaccessible products if you’re not actively working to remove barriers. I feel lucky to have worked somewhere that prioritises accessibility. Thanks to the work of some really brilliant people, it’s just part of the fabric at GDS. (For more on that work and those brilliant people, I can’t think of a better place to start than my colleague Ollie Byford’s talk on inclusive forms.)

I’m far from being an accessibility expert, but I can write about this because I’ve worked in an organisation where it’s always a central consideration. This shouldn’t be something to feel lucky about. It should be the default, but sadly we’re not there yet. Not even close.

Earlier this year, Domino’s pizza was successfully sued by a blind customer after he was unable to order food on their website or mobile app, despite using screen-reading software. And in a recent study carried out by disability equality charity, Scope, 50% of respondents said that they had given up on buying a product because the website, app or in-store machine had accessibility issues.

Legally, reputationally and most importantly, morally, we all have a duty to do better. To make sure our products and services are accessible to everyone. We can use design systems to help us on that journey, but they’re just one part of our toolkit.

In the end, it’s about committing to the cause – doing the work to make things accessible. Because accessible design is good design.

About the author

Amy is a content specialist and design systems advocate who’s spent the last 3 years working as a Senior Content Designer at the Government Digital Service.

In that time, she’s led the content strategy for the GOV.UK Design System, including a straightforward and inclusive approach to documentation.

In January, Amy will continue her work in this space, in her new role as Product Manager for Babylon Health’s design system, DNA.

More articles by Amy

Cambridge, Mass. : National Bureau of Economic Research, 2020

Cambridge, Mass. : National Bureau of Economic Research, 2020

Cambridge, Mass. : National Bureau of Economic Research, 2020

Cambridge, Mass. : National Bureau of Economic Research, 2020

Hoboken : Taylor & Francis, 2003.

London : Pluto Press, 2020.

New York : Springer, [2007]

Montreal : National Film Board of Canada, 1983.

Dewey Library - HX236.5.S6293 2020

Dewey Library - HV7561.H85 2020

Dewey Library - UA710.S135 2020

Dewey Library - HV6431.J674 2019

Dewey Library - HV9506.R53 2019

Online Resource

Dewey Library - JK2356.G76 2019

Dewey Library - JK2358.F5 D86 2019

Dewey Library - HV6250.4.W65 B69 2019

Dewey Library - HV6692.T39 L48 2019

Dewey Library - JK2071.H57 2019

Dewey Library - JC571.C78 2019

Dewey Library - JC599.A8 P53 2019

Dewey Library - UA668.I49 2019

Dewey Library - HV6626.2.S59 2019

Online Resource

Online Resource

Barker Library - UA23.H538 2019

Online Resource

Online Resource

Dewey Library - JC491.K255 2020

Dewey Library - JC423.U776 2019

Online Resource

Dewey Library - HV6432.B364 2019

Dewey Library - JK1896.W37 2019

Online Resource

Dewey Library - JC327.S79 2019

Online Resource

Online Resource

Dewey Library - JC261.A7 S74 2020

Dewey Library - UA23.B415 2019

Dewey Library - HV8079.B62 A44 2019

Dewey Library - JK468.I6 T76 2019