Comment by sambeau | Hacker News Reader

Comment by sambeau | Hacker News Reader

This smells to me of a team overthinking something so much that they land on something unintuitive. It smells of "over-fitting" — a solution way too specific when something general and flexible is needed.

Pressing shift three times is clever… but way too clever. Even if you stick a giant popup saying "hit shift three times to quickly exit" I'm not sure anyone in a panic will remember—loads of people don't even know which key is shift, especially when there's three buttons on a keyboard that look the same and only two are the same. I've come across people who always use shift-lock and did't realise you could use shift for anything. I'd be interested to know what UX tests they actually did, and who with.

If I was going down the press a key three times, I would have gone with pressing any key three times apart from the number keys (plus an info box when you enter the page—"hit any key 3 times to quickly exit to the weather"). Most people, I'm sure, would mash the spacebar in a panic but if they missed then it would still work.

What I would have preferred to test would be 'mashing'/chording — pressing more than one non-modifier key at the same time, so a user could just smash a load of keys at the same time in a panic.

Going to the Weather page is a great idea, though.

Hard disagree.

First of all. Not using escape key to escape is the standard for almost all applications since the 90s. Do you use escape to close the browser? A tab? your email client? No. All software converged on the idea that a close button was not a good idea, we are left with the actual button as a vestige.

Second of all, this software is designed for people in high stress situations where one of their main goals is to avoid detection, they will not only memorize the escape sequence, but they will likely have their finger on the shift key at all times.

I have my doubts about how sure you are that people in high stress situations will memorize the escape sequence for one website.

I think as devs we often think of our site or application as the center of the user's universe, but I don't think users memorize the minutia of our applications like we think / would hope.

Also, I actually worked with folks in abusive relationships at one time, their actions are not as predictable as you might hope.

> one website

I'm guessing gov.uk is hoping that this will become some kind of standard, at least for British resources.

> I think as devs we often think of our site or application as the center of the user's universe

Jakob's law is a thing but I actually think in the case of GDS they are in the fairly rare position of perhaps being able to justify the hubris you speak of slightly.

Not only are they directly or indirectly responsible for the UI of a frankly staggering number of online services, they are also one of the most influential bodies - perhaps in the world - when it comes to this sort of thing.

My only concern about setting a standard (beyond the usual process of setting a standard) it's that a standard for what exactly? All the other government sites that ... you don't need this key sequence on?

For the user I think that still means asking them to memorize something odd for a very limited use case that you won't think of visiting any other government site.

Ah, I misunderstood the scope of the tool.

I thought this was a tool that users specifically install in order to browse any content.

But instead it seems this is simply a feature so that users that browse gov.uk websites specifically can exit.

I think the OP's main point is "press shift 3 times" is a very uncommon and unintuitive keyboard shortcut. What do you disagree with?

> Not using escape key to escape is the standard for almost all applications since the 90s.

Really? I always hit "escape" when I get a popover on a website, and it often works.

Many TUI interfaces use it for "go back" or "exit" e.g. BIOS settings.

Using the Escape key to close dialogs is the standard for almost all desktop applications since the 90s, though.

^This. Closing an entire application with the possibility of losing important state? No, we don't want a button that does that (though a button combination is OK because that's less likely to be accidentally triggered). Closing an ephemeral popup that is distracting from the main page? Absolutely, `Esc` that.

> Second of all, this software is designed for people in high stress situations where one of their main goals is to avoid detection

FTFA: `It’s intended to be a safety tool. A way for people in unstable, potentially violent, domestic situations to quickly leave the page.`

This is the craziest part of this entire article to me. The UK Government needed to invent a whole design system that included an "ejection seat" button in case you're caught looking at UK Government websites?

Or does this button exist because one website in particular needed this feature?

Over design much?

It is specifically for websites offering help escaping domestic abuse or similar stuff:

> When to use this component > > Use the component on pages with sensitive information that could: > - put someone at risk of abuse or retaliation > - reveal someone’s plans to avoid or escape from harm > > For example, when a potential victim is using a service to help them leave a domestic abuser.

https://design-system.service.gov.uk/components/exit-this-pa...

IMO this seems quite well designed.

Why not exit the tab with Ctrl-W?

Hard disagree. I use it (and it works) so much that the rare occasion it doesn't is jarring.

Not to close a tab or entire application, no, but to unfocus a field, close a modal window or ad pop-up, back in something like a Typeform, etc.

I think it's useful to think about the same way you think about test specificity. Ie, of all the people in the world that hit this page, how many of them are going to need this feature and use it correctly vs. how many don't need this feature and accidentally use it. Using the Escape key is fantastic for "I needed this feature and it worked", which is probably 1 in 100,000 users of the page. It's terrible for "I accidentally used this feature I didn't know about" and that's the other 99,999.

All your other suggestions fail for this reason too - you need a high level of confidence the person really intended to escape. I for example would mash the space bar three times to scroll down.

Along these lines, in the GitHub discussion they show a graph of the number of times the button was pressed, bucketed by the platform the user was on, which is all utterly useless info.

It should be normalized as a percentage of page views at the very least.

They’re basically saying “hey we added a big red button and people press it sometimes”. The button could say “fire lasers at my cat” and some amount of people would press it (whether intentional or not).

Our browsers just need a boss-key.

Hitting shift 3 times happens just by holding the button down too long while typing caps sometimes too. I constantly have sticky keys coming up when inadvertently holding down shift and getting distracted while typing.

I've definitely triggered sticky-keys with my shift before, but I can't remember a time it was while typing - potentially while shift + arrow to highlight, though.

But it is one of those features that I turn off the second it annoys me 1 too many times.

Yeah anytime I'm on a fresh install of Windows, it seems to happen pretty quickly and then I turn it off.

There's also a giant red button you can click. That's the main route and it's pretty good for a paniced user who needs the solution right in front of them.

They keyboard shortcut is just gravy.

Yeah, this was my reaction... I wonder if they collect logs of how many people use the triple shift function. I do like GDS' focus on research and service design, but this feels slighly over-engineered from that space.

The logs are just noise without a way to prove the users’ intention to use the triple-shift feature for its intended purpose.

Maybe you could normalize it by listening for triple-shift presses on all pages on the site (not just sensitive ones) and calling that a baseline of accidental events.

But, how do we know that events in the baseline are truly accidental? What if users learned the behavior and tried using it on pages where it’s not implemented?

There’s just no good way to get analytics on this feature without interviewing users somehow.

Reply on news.ycombinator.com