Live data from Hacker News

How the Web Became Unreadable

backchannel.com

11–20 of 354 posts

Re: How the Web Became Unreadable

#11
I agree that the contrast on App Engine could be bumped up. But in their defense, most of the text/labels on that page are not meant to be read as if they are prose. As much I like Google's old-school blue aesthetic, the underlined hyperlinks in the sidebar were, IMHO, something that had to go away, as it's mostly obvious that every label, except for the bolded subheads, in the sidebar is a nav-item meant to be clicked.

In terms of font-size, I think it's a difficult tradeoff to make. I'm a bit impressed that the designers didn't try iconifying things and instead stuck with pure text. Again, as the admin page is not meant to be read like an article, the advantage of having small text is that at least it's possible to fit more things onto the page, in consistent locations, rather than forcing the user to scroll.

That's not to say that maybe some of the issue can be mitigated by reconsidering what needs to be on the page. I find myself being mostly OK with the typography on the App engine page because 95% of the content doesn't require my attention, and so making everything #000, as it was in the older version, has limited value to me.

I do prefer the older version's simple-table-list layout. I find Google Cloud Platform to be unusable because of its multi-column boxes-everywhere layout.

Re: How the Web Became Unreadable

#13
To make matters worse (i.e. take away even more contrast) users might be using f.lux, twilight or similar which will move the threshold of illegibility nearer.

(and there are valid reasons to use these, indeed I have them on phone and desktop).

Re: How the Web Became Unreadable

#14
post #6

We need some strong AI to strip away the ads, and extract only the main text, then format that text using the reader's preferences. OCR 2.0, if you will. Caveat: it may kill the advertisement industry.

Why stop there, implant it directly into the reader's brain.

Re: How the Web Became Unreadable

#15
post #9
post #6

We need some strong AI to strip away the ads, and extract only the main text, then format that text using the reader's preferences. OCR 2.0, if you will. Caveat: it may kill the advertisement industry.

killing the ad industry means killing the current web unless there is another revenue stream

It's that so? My experience says otherwise. For example, hn will stay, and so will the linked blog. They don't show ads.

Re: How the Web Became Unreadable

#16
Subjective though it may be, there are a lot of people who find black-on-white unpleasant to look at. I greatly prefer sites that don't use it, regardless of what they do with font and size.

This could just be a case of conflicting preferences, but I don't think it has to be. Clear contrasts are achievable without insisting on #000/#FFF web design.

- The most obvious improvement is simply keeping an eye on contrast. The article mentions using #333 over #000, but none of the unreadable samples provided are anywhere near that dark. #333 is approximately "light black" to the eye, and I suspect would give little trouble on a white background.

- Another trick is to vary one of background and text. Both Google and Apple violate this guideline by placing dark grey text on a light grey background. Again subjective, but offering a visual 'endpoint' seems to preserve clarity in a way that multiple shades of grey do not (perhaps it just encourages more careful design).

- Finally, we can choose colors which are not strictly grey. The HN background is a rather usable 244/244/235 - not far outside of light grey, but tan enough to provide more contrast than the pure brightness would imply. The vote/name options here are emphatically grey-on-grey, but it is at least a design choice to minimize their distractingness.

Re: How the Web Became Unreadable

#17
I have worried about this a lot, but basically now try to get inspiration and take example from webs that I consider highly readable. I also developed a set of internal rules; for instance, my maximum "black" is #333 (#333333) and my minimum "white" is #eee (#EEEEEE). Anything going further than these to the middle should have a real reason and not just because it's prettier.

Edit: #333 over a #eee (worst-case scenario with my guideliness) gives a 10.9:1 ratio, while #333 over #fff gives a 12.6:1 ratio, both totally acceptable. So I am not sure how the author concludes that #333 is not good enough and we should use #000;

Why should the people who is doing it right and pleasant change it because few others are using ridiculous colors as in the examples? It's like saying you shouldn't eat any sugar at all because many people is fat when you are perfectly healthy.

Re: How the Web Became Unreadable

#18
post #6

We need some strong AI to strip away the ads, and extract only the main text, then format that text using the reader's preferences. OCR 2.0, if you will. Caveat: it may kill the advertisement industry.

That doesn't require strong AI, firefox reader view achieves that without strong AI. It even preserves article images nicely.

Re: How the Web Became Unreadable

#19
post #9

Earlier quoted context omitted.

killing the ad industry means killing the current web unless there is another revenue stream

It's that so? My experience says otherwise. For example, hn will stay, and so will the linked blog. They don't show ads.

HN is a meta-site. Without ads, there will be dramatically less content to link to.

Re: How the Web Became Unreadable

#20
post #6

We need some strong AI to strip away the ads, and extract only the main text, then format that text using the reader's preferences. OCR 2.0, if you will. Caveat: it may kill the advertisement industry.

That doesn't require strong AI, firefox reader view achieves that without strong AI. It even preserves article images nicely.

Reader view isn't perfect yet, which is probably why (useful as it is) most people don't use it.

Also, as more readers use it, the ad industry will try to circumvent it, e.g, by tricks like turning part of the text into images, et cetera.

Eventually, you will probably be better off by OCR'ing the rendered webpage (i.e., as the user sees it), and using AI to only keep the relevant text and images.

Post reply on HN