# Shall we make it explicit site policy that screenshots of text ≠ readable text?

As a helpful example only, this otherwise great answer to Is Starship planned to fly directly to the ISS without first stage? Is it even possible? relies heavily on the readers ability to read screenshots of tweets. The following ~120 words from four tweets are available only to folks who can interpret screenshots of text.

SSTO = Not gonna happen. People need to stop peeing their pants for an SSTO. YES, getting to orbit is possible with most modern rockets with a single stage. But doing any actual work and ESPECIALLY having margins to come back and land is no. It's just not beneficial.

Exactly. We’re on the wrong planet for SSTO. No problem on Mars or any of the moons.

Correct me if I'm wrong, but didn't @elonmusk state a few times that StarShip cannot do SSTO on Earth? Did I miss an update? Are the raptors more powerful than earlier thought?

It technically could, but wouldn’t have enough mass margin for a heat shield, landing propellant or legs, so not reusable

Since I often struggle reading non-English websites where critical information is displayed as images of logograms (as opposed to an alphabet that one might learn in a finite amount of time) instead of selectable text that can be copy/pasted into google translate, I have some small degree of understanding how frustrating this might be to folks using screen readers.

Question: Shall we make it explicit site policy that screenshots of text ≠ readable text? If we have a well-received answer that says simply that any text information that's a necessary part of an answer be displayed as text, it can be added to our Community Policy Repository.

I'd say keep it narrow; no reason for an absolutely ban on screenshots that contain text, but as a minimum requirement if one really wants them in a post one must do both of the following:

1. Also past the text as a block quote for folks who don't read pictures and for search engines.
2. Replace [enter image description here] with the URL of the tweet, or an statement that it's the same text that's block quoted, or if it's short, the text itself.

An answer could recommend best practices, but primarily the goal is to make sure any text that's a necessary part of the answer is present as plain text for screen readers and search engines, and the associated images contain something more helpful than [enter image description here].

• We have no site-policy or post-content or images tag.
– uhoh
Jul 16, 2021 at 22:49
• I have long complained about StackExchange / StackOverflow lack of accessibility, particularly with respect to images. Images of text are usually by those who want us to do their homework for them. They don't even take the time to write the homework question, instead snapping a picture of the textbook and posting that. As far as I'm concerned, those questions should earn a GTFO close vote. (We don't have a GTFO close vote reason.) Jul 23, 2021 at 15:21
• One key problem with the [enter image description here] is that the image description does not qualify as alt text of the kind that screen readers can use. It is instead the text that SE/SO uses when the image can't be rendered. SE/SO does not support alt text. I've asked, and apparently this support might be forthcoming when the Sun becomes a red giant. Or maybe later. I still do try to provide image descriptions. Jul 23, 2021 at 15:28

I don't think we can have a blanket policy because screenshots of text are useful at times. When screenshots or imagery is necessary, ask the questioner or answerer to add an image description that is better than enter image description here. Since a good picture is worth 1000 words, that replacement for enter image description here had better be wordy.

For the vast majority of times where screenshots are not necessary (e.g., the linked answer), don't upvote those questions and answers. Downvote them instead. For questions, downvote and then vote to close.

I long ago became sensitized to accessibility issues because I have had the good fortune of working with an absolutely brilliant NASA mathematician. We've worked together on multiple projects over the course of three decades. Sometimes it's been years between assignments, but I have always enjoyed working with him. He is totally blind. Lack of accessibility is a key challenge for him.

Lack of accessibility is an even bigger problem for visually impaired people who are not absolutely brilliant. Only 44 percent of people who are blind or visually impaired are employed, compared with 79 percent of those without disabilities. Accessibility is a touchy issue for me.

Not that I'm an active member with bazillions of reputations and so forth, but I agree that images-as-text is quite annoying. However, I think there are some exceptions where, though annoying, it makes sense to not be too picky.

1. Text isn't particularly necessary, or doesn't make sense without the image anyways.

For example, I've posted a number of diagrams where "this doohickey goes..." "...into this slot" with arrows and so forth. Without the image, the text is meaningless, so it doesn't help people with screen readers (probably, since they're probably blind enough the diagram isn't legible). I've described what's on the diagram in the text of my post, so hopefully the diagram is still useful, and it's a bit overkill to create 30 diagrams written in common languages (and I'd probably screw up the translation anyways).

That said, it's possible to create diagrams with only numbers, then write text statements below the image to support the numbers. Then the statements could be translated as long as the person is sighted and knows Arabic numerals. (This also has the side benefit of less clutter in the example image.)

Example
To create bold text, do this:

"My Example", taken from www.example.com under CCxxx license. Doqath, 1952

1. Highlight the text you wish to bold.
2. Then click the "B" icon to automagically add the "bold" markup.
2. The text is completely incidental.

If I post a picture of a spaceship and ask "what's the thing on the side of the tube for?", the fact that someone captioned the image "SpaceX to launch new rocket" isn't helpful to translate.

3. There's way too much text to transcribe, and the original information wasn't available in a copy/paste format.

Not sure a good example for this site, but there are plenty of places in computers I might take screenshots of my settings windows so people can help me better. Given the amount of raw data I'm conveying, it's likely not worth the effort of manually transcribing each and every field (and I might well make a typo that makes it hard to fix my problem). In these cases, exported log files are better, but not always easily available.

Another example would be a photograph of a book or manual. I might be able to provide you with the information you want, but I don't feel like transcribing 3 pages of technical details and cropping images to make a web-friendly variant. Sure, it would be a better answer if I did those things, but we're not getting paid by the hour to post on the site.

Still, I think it's reasonable in most cases to transcribe the relevant paragraph from physical media since most answers requiring 3 pages of tech manual are probably out of scope for the site.

One thing I would not recommend is circumventing the image description field with a URL. That field is for describing the image, not linking to other things. It's easy enough to either make the image into a link, or post a link right under (or above) the image in a caption / attribution field.

Example image links to www.example.com when clicked, and has this neat caption.

Note that by default an image links to a full resolution version of itself. For large images, the site scales them to fit into a certain size to keep the interface clean, then you can click the image to see it zoomed in. Replacing that link with a link to another website is possible, but makes it more difficult to see the zoomed in image.

For small images, re-linking isn't a problem, since there's no need to zoom in, but it might not be obvious the image links to Twitter or wherever. With the same space and effort it takes to link the image to another site then add a note explaining this, you could simply add a caption with an obvious link.

As a parting thought, I'd like to point out one good reason to post images-as-text is to create more permanent evidence a certain text actually existed as claimed. People delete posts, change signage, and generally don't care about your SE post's authenticity. While most of the time you don't really need to prove your source, sometimes it's a good idea for posterity. Likewise, a picture of that 1847 dictionary showing how the word was defined back then makes us feel much better than some typed text claiming that's how it was defined.

• Thank you for your answer! Do you think we should "...make sure any text that's a necessary part of the answer is present as plain text for screen readers and search engines, and the associated images contain something more helpful than [enter image description here]"?
– uhoh
Jul 18, 2021 at 12:47
• @uhoh: I think it should at least be strongly encouraged, and I don't see any reason it can't be an official policy. I think you're less likely to get descriptions on images though, as many won't see it as helpful enough to warrant their time. Even though I always write a description, I'm still not sure how helpful they really are, as I've never gotten any kind of feedback from someone with a screen reader. Jul 18, 2021 at 12:57
• "Another example would be a photograph of a book or manual. I might be able to provide you with the information you want, but I don't feel like transcribing 3 pages of technical details and cropping images to make a web-friendly variant. Sure, it would be a better answer if I did those things, but we're not getting paid by the hour to post on the site." Agree with this. I'm pretty sure I've answered a question by scanning a page from Sutton, in a case where typing in all the equations with my primitive MathJax skills just wasn't going to happen. Jul 18, 2021 at 18:35

Many PDFs already have copyable text; the older ones through optical character recognition, and the newer ones natively. Most of the useful information on web pages is native text that is copyable. It's actually more work to take a screenshot, save the file, and then upload the file to the question than to simply cut and paste the text!

I answer a lot of questions about the Mercury to Apollo era, for which documents are usually OCR'd instead of native text. The conversion is often pretty bad, with mangled characters and spaces in the wrong places. I do my best to fix the text up -- or retype it -- and include it as a quote. Yes, there are some times when a larger section is helpful, yet too long and not worth to transcribe or fix up; in that case, transcribe the minimal amount of text that supports your answer.

I agree that text in drawings, graphs, and tables are often not worth transcribing. However, there still should be your own text briefly explaining the purpose of the figure, and pointing out what we are supposed to look for in the figure.

So maybe not a site policy, but certainly an expected behavior in most cases.

• "It's actually more work to take a screenshot, save the file, and then upload the file to the question than to simply cut and paste the text!" Not if you use ksnip. That can directly upload the area of the screen you outlined to imgur, and you don't have to save a local copy. Shift-Alt-R, drag the cursor, Ctrl-U and you're done. Jul 24, 2021 at 0:13