How To Get Your Newsroom To Care About Alt Text And Accessibility

6:00PM Aug 25, 2023

Speakers:

Keywords:

alt text

graphics

text

screen readers

newsroom

barbie

people

alt

visual

case

article

write

work

image

photo

editors

jamie

share

new york times

story

So I think we'll just give it maybe until about two minutes. At about 232 And then we'll get started

how's it going?

Could be

a great thing that when you're

the title of your session is, you've seen it once and you know what it is and I corrected them and I was like, Okay, I know where I'm going instead of like, all these are complicated

Oh, wow. That's exactly what he talks about. Being concise.

Plain language.

Like I'm so proud of my Goldstrike.

Started. Let's do it. Okay, well, hi, everybody. Thank you so much for joining us today. At our talk about how to get your newsrooms to embrace all texts and accessibility at large. We are going to be talking specifically about all texts, but actually a lot of the lessons that we're going to be sharing that we learned at the New York Times can also be applied to larger accessibility initiatives. So we're going to do some introductions and then we will go ahead and get started. So my name is Bina Ravindran. I'm a senior editor at the New York Times. I work on a team that trains the newsroom on everything from new tools to culture change and lots of different initiatives. And our initiative to get our newsroom to add alt text on photos and illustrations was something that I helped lead in my team helped lead as a visual description of what I look like I am a South Asian woman and I have short black hair. Through this rollout in our newsroom, I worked really closely with my colleague Jamie Tanner, who will introduce herself, but focused a lot on what this looks like on our graphics.

So hi, my name is Jamie Tanner pronouns she her visual description of a white woman with long brown hair. We're an All Black and I am a visuals accessibility editor at the New York Times. I work with our visual reporters, but largely graphics and interactives and I'm there to help make sure that our work is as accessible as possible that sometimes takes the form of helping with our tooling to make sure that we're able to do that through training and guidance and also editing and auditing stories that we put out there to ensure that they're as accessible as possible. And we also have here with us today Swathi netic Kumar from the American Council of the Blind, a group that advocates for equality of blind and low vision people. They're also one of the groups that we approached early on to ask for feedback on our alt text initiative. And they also helped us get connected with screenreader users over the units all over the country who could give us some feedback on some of our early published alt texts at the New York Times and Swathi was also one of our very first beta testers. So super glad he could be with you today. I'll let you introduce yourself. Yeah, hi

everyone. My name is swappa Nanda Kumar, I am from the American Council of the Blind. I am their advocacy and outreach specialist. I use she her pronouns, and I am a South Asian woman with a long black hair and a black suit. Thanks for having me. It'd be good to be here.

Yeah, we're really really excited to be talking about this and we're gonna have some fun and it's going to be interactive. So I hope you all are ready to, to, you know, talk and share. Okay, I think yeah, we're kicking it off with a poll. Okay. So, with a show of hands, could you please raise your hand if you yourself have ever written a piece of all text that has been published that includes in your newsroom or on social media? Okay. Okay, great. Second question. Um, can you please raise your hand if you know that your newsroom or the place that you work in consistently publishes all text on it's photos, illustrations and visuals. And if you don't know if you're not sure you can like put up a peace sign or something like two fingers. Okay. So we have like, we have some hands we have some like, you know, not really sure we have some peace signs. Um, so it's great because we're going to be talking about all of this in our session. Today we're going to be talking about some tips and some lessons that we learned as we rolled out all texts. Spoiler alert, we're gonna get into this but it took us three years from start to finish depending on what point of the timeline that you use to fully add alt text to all of our photos and our illustrations and our graphics. And it's a process where we learned a lot, we're still evolving. We made mistakes along the way, and we're going to talk about some of the things that we learned. So hope you all are ready and and Yeah, down to kind of talk more about accessibility.

Cool. All right. So this is where we're going today. We'll start with just what alt text is for those of you who are new. We'll talk a little bit about our best practices and share some examples of published alt texts that we have at the times we'll talk about how we got there. And some tips on maybe getting your newsroom to embrace it as well. And then yeah, as Peter mentioned, we'll we'll talk a little bit about the beta testing that we did and share some some feedback that we got there. We'll talk about what we heard within the newsroom and share some some feedback there. And then we'll talk about some common mistakes that we've noticed as we've been auditing, the alt texts that we've been publishing for the last few months. And then we'll wrap it up hopefully with just a conversation about building a culture of accessibility in newsrooms. Okay, so to start, what is Alt Text? Very simply it is a visual description of an image. It's, you know, a text description that tells you exactly what it is that you see in an image. And that text description is used or can be accessed by screen readers which blind and low vision people use when they're reading the news and so we have an image here, and an example of the alt text for this image, which reads, a nurse wearing a blue protective smock and white gloves administers a vaccine into the upper arm of a masked patient. And we also have a caption here because I think this is a question that comes up all the time. How is that different? A caption adds extra context that might be helpful to that image. And so whether you're you see that image whether you hear a description of that image, whether you're reading a description of that image, using a braille reader, you know, for all of the scenarios, a caption is used to add some additional context there and swap that jump in if there's anything that you would add to that that I missed like two.

I would just add that again, like this just is grabbing what the picture is. So kind of just like how you see it, just say what you see. So

definitely, and we have an example too, of what it sounds like when you're using a screen reader. I'll just hit play if you want to,

actually so this is an example of what it sounds like when you don't have all text. So this is a screen recording that we took very early in our process of researching alt texts, because we wanted to understand what it does to the experience when you don't have alt text on an image. So some things to listen for here. This is a real demo of how a screen reader works in a screen reader scans all of the assets on a web page and reads them aloud. In this case, specifically, the caption is pretty contextual. So just like why don't we just take a beat and like kind of listen to what this sounds like and then we could talk about it.

36 years we'll start this week. Wager Lisa saying easy on me, followed by a new album on November 19. She announced on Wednesday. This airs I feel like I've finally found my heels again. I don't say that the statements announcing airport damage. You are currently on a mission to end so. Mario 80 on slash riders. You

okay, so what did you all think about that experience? What are what are some things you can take away? From that? Feel free to shout it out. Yeah.

Like, but if we're sorta hearing impaired, I wouldn't be able to like differentiate all the words.

Yeah. Yeah. And just to say that for anyone who couldn't hear, it sounds like garbled and maybe you can't differentiate between all the different words. Anybody else have another takeaway to share? And yeah, it's not descriptive. We actually have no idea that that is Adele in the photo. Because the caption is a is literally a quote from Adele, which makes for a perfectly great caption but not a great or very accessible experience. swappa is somebody who uses a screen reader all the time. What other I don't know what are their takeaways do you have when hearing an article that doesn't have all text on it?

You really have six bands of the article like I know, but I pick on the article I can tell that Dell has, you know, six albums and this one is a new kind of album coming out or Mysskin coming out, but I wouldn't know like just that image so don't know what the image is up or what what kind of add your article so just reduce my experience of the article from like, just general experience of news journalism in general.

Absolutely. Yeah. So a lot of information but not a lot. of understanding of what the visual is contributing there. Yeah, I'm gonna go to the next one, Jamie. Okay, so then for one of our first pieces of published all text, we actually published it on another Adele story, which fell very full circle, but here's something that you can listen to here. So this story, and this image does have alt text on it. So we've kind of wanted to do a compare and contrast. So maybe you can hear any differences that we experience with alt text provides here if you want to hit play Jamie

Lewis six years he was a divorce and moving on character and misery with our show Aussie. You are currently on social media shamba wearing the blackout and accompanying her knees to her chest in front backdrop, in which the big boys who are arguing jeez, imploring query writing shouting is rightfully at the center of 30.

So obviously there we also picked a pretty flowery caption to demonstrate the power of really just adding one sentence of alt text here and a pretty basic description of Adele and kind of the pose that she's in. I hope that you know you all can see kind of, you know, just what that one sentence does to improving accessibility on this.

And all text and image descriptions can look a little bit different for graphics. So we pulled an example of a story from a story that just published this week. Just to give an example of what that might look like. In this story, there was lots of looping videos gifts of like simulations of near misses on airplanes, which was very disturbing story, but hopefully it gives you an example of what that experience is like for some graphics do.

We also have the words International Airport. Present was preparing to take off on the table they decided to do it in a positive position by sentence. You are currently on the test. Graphing group. You are currently in Louisiana aviation to Texas I have words is something I saw a year ago for learning the updates offered the same way causing the first day school sharpies here. No animation sounds like Chinese vertical. Lines are really horrible considering that graphic group nowadays isn't.

Alright, I'll pause it there. So some interesting things to note here. Maybe one we've like wrapped it in a group. And so a lot of times when we have graphics, there's sources, there's notes, there's lots of other like pieces of text that are really tied to the graphic, but you wouldn't necessarily know and so we've added something here to build that relationship to communicate that relationship that they are all part of one group. And then here it's a simulation like describing what's going on in the text and so really would take away lots of important information I feel from the story if you're not communicating or translating what's going on in that video graphic. Anything else you would know there's a lot there to feel free to jump in and interrupt us

no I think I think that like it goes pretty pretty well covered it. Like I do think that and we do like yeah, give the caption cuz how that how about alternate data on that but like graphic just uses simple and like show like what's going on. So it's gonna read, read read it out. So totally, is

not what I want. There we go. Okay, so that's some examples. And here is just some tips that we pulled out from our own best practices. You wanna start with photos?

Yes. So our photo and illustration guidance, feel free to take photos but we're also going to, there is a PDF on the session page and I also have a bitly later in the presentation where actually we have like even fuller guidance. This is just like the high level if you're going to start doing anything today, what you can start to do but there there is a lot more that we're going to share and kind of open source but kind of some some big things to take away here. Really limiting to one to two sentences as you saw in our photo examples so far. This is meant to be succinct a simple visual description include physical attributes of subjects, and then describing things like colors, pattern, action, movement, those things are really important location or important background detail. Oh, and then also including the names of visual subjects. We're going to get into that and show why that's important in an example later, and punctuation because as you can hear the screen reader does know when to pause it does know you know when there's a period or something at the end of the sentence to

Yeah, for graphics. The same is true. You really do want to keep it as concise as possible, which can be really challenging for graphics, I think. So do your best sometimes it's gonna be a little bit longer than a photo. That's okay. Incomplete Sentences are okay, I think you see that also with graphics. Lots of times. There's just lots of data that you're communicating and it's a lot and it's okay to pause. I think that can be helpful. And then three big things that I think we really want to hit on each time is to one describe the graphic, is it a map is it a video is it a diagram, really at that level is fine saying charts okay, we usually don't get too into the weeds on to marry Mecco or whatever, you know, really descriptive chart name it is. So what kind of graphic it is. Communicate the data that's being used. And a big point is summarize the the main point of the graphic or what information is being communicated. Don't just say it's a map, what is it a map of and finally, we use annotations a lot to kind of highlight things sort of pull out important information. Use that whenever you can. I think that's that's always there to serve you too.

And then I can go through the things to avoid, you'll see that it's just generally avoiding assumptions, which is always what we're trying to do as journalists. But specifically in our alt text, it is avoiding assumptions about race, gender, identity, age, really any other kinds of identity. The same rules go if you don't know it in your reporting. It should not go in the alt text and there's even some other best practices that we have around identity use in all text, which we're gonna get to a little bit later. Same goes for data or trends, no assumptions, please on that. And then the last thing is avoiding repeating information. You know, folks we're using screen readers also have access to the entire story, including the headline, the sub headline, the caption, so you know, we don't need to make it super redundant. And just thinking about what your alt text is doing in the architecture of the story is really helpful there. swappa Any other big high level points that we missed there?

Yeah, just like capture which was relevant to story and not like the previous details. So like, you said, like, don't know assume rates like, if recently, like it's important, it is important to the stories and like face a diverse crowd or like, make it general but also do not assume but also an abuse. You're going to use it like by slate when the story calls for it. So

great point, and we're gonna get to that an example that actually illustrates that in a couple of slides. Swathi you also made a great point when we were chatting earlier about the length of the alt text and and how it should be matched the complexity of the photo. You want to say more about that? Yeah, I

think so. So for all texts, shorter is often better, but also like make sure to capture everything everything relevant to start picture. So like I feel like been one of our earlier testing, like test sessions we, we have a picture of a painting and the painting can't be described simply in one in one to two lines. So you really want to like capture all new monster capture, like colors and details. So kind of make it like the it can infer length, like based on what you describe. So so it's like, again, like just make it concise but also make it like doesn't have to be confined to have to have to be like short and sweet. So

great points. Thank you.

All right. Oh, yeah, this is just another just to get into graphics a little more because they're very complicated. I think this is a template that we adapted from Amy Cecil's article writing all texts on data visualization that I share a lot because I think if you're stuck, it's a great place to start and again, kind of hits on the three pieces of information that we usually want. Those being the graphic type, what type of data and why you've included this graphic of what the most important piece of information is. And so I'll just read the alt text that we have here a map showing the locations of abortion providers that have closed, stopped offering abortion services or opened new opened a new location. Most of the clinics shown have closed or stopped offering abortions, or in the 14 states that ban abortion and most new abortion providers shown are in states that haven't so I think you know, this was published as pieces as a part of this article. And I think they did a really great job because the the graphic is pretty complex. There's a there's a color that we're using to label each of the states there's different points on top of each of those that are labeling that a separate piece of data. You could get into the weeds of describing all of those color differences and all of the shapes that we're using here, but this does a nice job of kind of concisely describing what it is we're looking for. And the main takeaway of this graphic that you're supposed to get as you're reading through the article. So now we have some just some examples of places where we use this guidance. Starting with another graphic here. We're not necessarily using that, that template, but following some of the same ideas. So this was a flow diagram that appeared in this article titled we hit the debt limit what happens now and so the article started with a pretty complex example or a pretty complex flow diagram and throughout the article, we would highlight different pathways. And so I'll just read out the alt text that was written for this article. A diagram highlights one of three strategies that the Treasury Department may try in order to avoid default if Congress lets X date pass without raising the debt limit, prioritizing interest payments. I love this concisely they did this just so artful. But you know, I think in some cases when you're writing Alt Text for Graphics, it can be really helpful to use the structure to help you kind of shape a description of what you're seeing in the graphic here. They kind of took that apart and took a more narrative structure to get to the same point and I see a question, go for it.

Yeah, great question. So we take a different approach. Maybe you could speak to that too, because

you want to take it first. Well, I'll

talk about in, in graphics, I think the approach that we've we've taken No, it's good to ask now too. A lot of times as I don't been as you've visited different desks, you first started by talking with the desks and understanding their process. In the work of photo and illustrations. I think it's often the case that the first editors or the story editor is writing the alt text. In the case of graphics. Graphics editors are often writing in and sometimes story editors are writing it as well. But it's very often the case that the graphics editor kind of knows the piece best and knows the image best. And so we found in our case, that works. So it does differ between desks, by and large advocate story editors, and we've seen the most success with that right.

And we'll get into that in a couple slides as well. Like just in your newsroom, how you can think about workflow and who's responsible for doing it.

Graphics and when you're using it is the maker.

Yes. That's a good point. The graphic creator

graphics editors.

Very good point. And I see a question back there to

potentially save us one of the three strategies if you're using student loan strategies. So I'm presuming your rightful place like that one.

Yeah. Yeah, well, here, the the price the let me pause the strategy that they're they're using one of the three. Wait, am I reading that right is prioritizing interest. But you're right that like within like this sits within the context of an article in between each one of these highlighted paths. Exactly. And that's like something to think about too. When the graphic leaves the context of the story and makes its way to social media. For example, there's sometimes some like retranslation that needs to happen there. And it's also true that we have the ability to kind of expose or hide some of the text elements that you see on the page here as well. And once they make their way to social, they get flattened out. And so you want to also kind of reassess that as it leaves the page. Here there's a lot of work being done by like visual arrows. And you see that all the time with like network diagrams and flow diagrams like this. And so it's like a little bit of extra work to kind of describe what that journey is individuals, right. So I'll move on to another photo exam. Yeah,

and we're just going to go through a few examples to sort of show the range of visuals that we at the Times publish. Just wanted to show this, to give a sense of like all text is going on all of these types of pieces just to give a sense of the range. But also because we're gonna get into like a pretty specific case study on how we rolled this out how we rolled out this training. And these are like parts of the reasons why it took a really long time and it took a lot of work. And as Jamie alluded to, like a lot of conversations about workflow and listening and understanding what was already on people's plates, and how alt texts can be prioritized how it can become like a required part of our news report. So also to kind of illustrate some of those pieces in here. So for news photography, all text would read, in this case, President Biden in a blue suit and tie speaking and gesturing behind a lectern. This was literally the published alt text for this image. And the caption you can see is very contextual, the Biden administration argued that tech companies is content moderation decisions were free expression protected by the First Amendment. In this slide, I also wanted to demonstrate we write alt text on our illustrations to even when they don't run with captions, the alt text is required on them. So here the alt text reads and illustration of a computer screen showing a virtual class with eight squares. One square has a teacher in it and each of the seven squares has a student. Now we're gonna get to a really fun part, which is you all are going to write some alt texts. Let's do it. And we're gonna start out with a really fun steal. Who here saw Barbie and knows this image? Great. So if I don't want to make any assumptions, but if you can see the image, what we're going to do is write a sentence or two of alt text for it. You can also collaborate with the person that you're sitting next to, if you have something to write with, like, maybe jot down a sentence or two, and maybe we'll we'll take like a minute or so and do this. Then we're going to come back together and we're going to ask brave souls to share their alt texts and then we're gonna like give you feedback on it in real time, but we're going to be really nice about it because we're all learning and I know some of us might be doing this for the first time. So I promise it won't be scary. But can we just take a minute and think about write down jot down some alt text again, it can be collaborative, as well. So we'll come back together in a minute.

I have still not seen Barbie. And I'm so sad

if you can if you have

something to look forward to

come on in. We're quiet because we're making everyone write all text.

We are currently writing all text for the image on the screen

are we on time?

Three o'clock. All right. Y'all just want to like look up when you're done, or something that'll help me figure out when to come back together

some, some eye contact, eye contact

All right. Well, why don't we bring it back together? And you can jot down your final, final thoughts. Do we have a brave soul or a brave soul pair to share your all text? All right, I'm gonna I'm gonna bring you the microphone.

Okay, can run.

Thank you my first time on a microphone though and I am excited. Okay, so I've got Margot Robbie as Bobby sings or last. I'm not sure what she's doing. Haven't seen the movie while driving her pink Barbie car. Ryan Gosling sings all us along in the backseat wearing a pink t shirt. There is a rainbow in the sky.

Community snaps for our first brave soul. Okay, can I be back panel in the front? Give us any early feedback on that

swapper any thoughts on length type of description anything that pops out to you that good jobs helped her

Do I have good Yeah, it's like like the cyclic close debridement if could attend. Yeah, good. Good. Nice. Nice work. Yeah.

Sorry, I just asked me a question. I this might be a really dumb question. How relevant is it to describe things that are visual when you're describing for people who have visual impairment and I'm thinking about things like colors, light and dark? I presume some people who have visual impairment can still sense different colors and things like that. Is that a reasonable statement?

Yeah. So again, like for people who are blind is blind spines is a spectrum so you can you up to your grade from like, totally blind and not being able to like empty colors and like people that can see like, like most objects in front of them, so I still think it's relevant. Especially because like this one Barbie is pink or pink so but I mean again, like it was like just hundreds of story so are what the picture we have for now particular story so I mean, it's worth it we're doing it is you need everything. It needs that so

yeah, that's That's great. So I thought and just one thing to add there because we got that question in our newsroom too, is what we're doing with this alt text is we are writing this for a really large audience. Like there are a lot of people who use screen readers for a lot of different reasons. And so I think going in with an assumption that the person or the people that we're writing for are like a specific type of person is actually not true. There are millions of people who use this technology across the Internet and yeah, so we're just like thinking broad about, you know, kind of meeting needs of a lot of different people is maybe another way to look at it. Let's do another Is it a question or do you want to share your example? Oh,

okay. So because Margot Robbie's in the caption, I didn't say Margot Robbie, I didn't want to double up that information. Okay, so I said actors playing Barbie happily driving her pink Barbie car with happy Ken riding in the back, driving away from a rainbow and through a desert

was good. Again like the you don't necessarily need to like have Margaret magrabi in there. If you didn't know you didn't know if Margot Robbie or like even know Kenworth rank up Winco. I mean, yeah, it's just so good. Yeah, just doesn't need you don't necessarily need to have the names of the people in there. So

yeah, very nice job. And one thing I wanted to point out in both of your alt text is the the verbs or the description of expression or how people might be feeling so we had a lot of discussion in our newsroom on this and I'm curious what what others think in the room but kind of where we landed is even when it's a movie character and yeah, they like look super happy because that's not something that we know through our reporting. We at the Times opt to like not say that and instead go with verbs like laughing or singing, but it's also hard in this case, because yeah, I saw the movie and I'm like, pretty sure that they're singing here and yeah, that's right. Yeah. Are they? Yes, it's the point where she's screaming cuz yeah.

Yeah, so see, even in our room right here, right? Now, even in our room of people who've seen Barbie there's even there's even some confusion around what the characters are doing. So what I would do and kind of what we train on in our newsroom is riding around that as much as we can. So if we if this was like a different image and we could very clearly say that she was smiling, Margot Robbie, we would do that, in this case where we're not really sure if they're screaming or they're singing, we might just opt for saying like they're driving, and it's pink and describing what we know in the image is like probably what we would opt to do. Yes. handout

so this is an entertainment handout images, so your staff photographer, you having your staff divers write the alt text, or or no they're in their field or they put just giving caption or they can they give alt text?

That's a great question. I think it depends on your particular newsroom in that case, or for the New York Times since this is a handout like we we have given this training to our photo desks, so they're in the loop on it. But as we were saying the responsibility for writing the alt text is on our editors. If you're in a newsroom, I think with like a full staff of photographers, which is like not the New York Times the situation. I think ostensibly like you could give training to your photographers. So that this information is also in the moment that they are like filing the photo.

When photographers in the field file their stories they input some submitted data comes from like the photography equipment, but they do identify some people so they don't have to write the all caption but there's elements there that the photo editors are gonna use to build the that are either on the story that the reporter is reporting or on the metadata that the photographer is inputting. But the photographer doesn't have to write up the alt text. Yeah. Yes.

Would it be helpful I should say my name is Molly. I have I actually said give me Barbie boardroom Barbie hair. My hair is blue and pink. And my pronouns are she her if that helps. Also, I'm a white woman millennial aged anyway. So is it okay if I say mouths open? Is that too much information? Because you said driving?

No. That's fine. What do you think? Yeah, you can find to Al's gonna I'm gonna mention that like, if you know that. You can see what the mouth open or you're like, No, I bugged out. Like you can say that too. So yeah.

I think we're gonna move on because we also have a graphic that we're gonna write all text for to

do quickly. I did see what of course I saw your hand raise for a while I told her. Oh, yeah.

I was just wondering about I understand like, there's like different types of pictures and images. But like for movie or other cultural media where a lot of the frame is very important. Would it make sense to describe the background as well because I was only thinking about the foreground and an alt text, but I also get that that's part of movies. So just wondering what I guess best practice advice would be for that.

Yes, actually, Jamie if you want to click will show what our suggested alt text was for this, so I'll read it aloud. In a steal from Barbie Margot Robbie as Barbie dressed in head to toe pink drives a pink convertible with Ryan Gosling as can also in pink in the backseat. They're driving through the desert with a sign reading Barbie land behind them. So we do have some in the foreground, some in the background. We opted for putting in names here but I totally take swath as point of like that that might be evidence in this particular situation, but we actually have larger guidance in our newsroom about why it's important to identify the names of visual subjects in the alt text, which we're gonna get to an example that kind of demonstrates that in a second. Yeah.

But I would say like a big takeaway I feel like from this exercise, which we do in the newsroom also, is that if we had 100 people in this room, everyone would write something slightly different and they could all work very well, like these are, you know, I think that some of the fun of reading this all text now that we're writing it pretty consistently is that you kind of like see different perspectives and you start to get to know people's voice and the way that they write this all text and different elements come to surface along the way. Yeah. Good question. Let's go.

Since you might be getting there so I don't want to Yeah, we are we are getting that added to that. So that was it. Okay, if we we have a slide that

goes like so I will shut up.

Okay, if we don't if we don't answer it please feel free to ask more. Okay.

All right. This one's harder. So let's see, how are we doing on time we can we can, we can try to go for it. So I wanted to give a descriptor, an example of a graphic as well for us to try and I saw the Barbie example you had I was like perfect. I have a Barbie in Heimer weekend box office charts that we can show. So why don't we try the same exercise with this and we'll we'll try to give let's two minutes, let's say yeah. And yeah, let's see where you get. And just I guess like as to answer a question that might come up ahead of time. If you were navigating this graphic within a news story. On the screen reader, you would get the head and the sub head and the caption so you know, keep it focused. on just the chart.

Lots of great questions.

All right. I'll just ask even if we're still writing, maybe if anyone has anything ready and is willing to share, just raise a hand and being a compiler runner again.

Well, I hadn't finished writing mine. But um, and I don't, but I was going to do something like a bar graph shows Barbie head 52% of the weekend and Oppenheimer had 26% And then I read something that said, when you're showing a graph, you don't always have to put in all the information but just a few key pieces so I was maybe gonna then maybe mentioned Star Wars and infinity war and then just end by saying Avengers endgame had 89% of that weekend total. I don't know. That's what I was gonna do.

Totally. I think you're like grappling with a lot of the questions that we all grapple with every time we're writing all textbook graphics. So I think that's, that's a great example. So we speak to that one. Yeah, I think you you first mentioned mentioning the 52% and the 26%. I think, in our case very often, you know, for so much of this I think like I said in the beginning when we're talking about the templates, so much of the conversation I think around all text and graphics is around what you want to communicate. What's the most important takeaway for this? I know from the story here, and from the title that really the biggest takeaway from this graphic was that in the four biggest box office weekends, it's usually been one really dominant movie and in this one movie in this one weekend, even though there was two movies that didn't actually make as much as these other ones have, historically, the combination of both of these big box office hits happening at the exact same time to get over that 300 million mark. So that's, you know, I think something that one again, I think it speaks to why having the the graphics creator often is so powerful, because they know well, what they're, they're hoping to kind of communicate in that graphic and what the most kind of critical information to communicate is, but also that there's just so many stories that you could tell in the graphic and lots of information and I think the point that you raised is a great one that like you could list all of that information out and and oftentimes for the altex We tried to kind of like take away yeah, that biggest, most meaningful piece of information and communicating that which, again, a graphics graphics creator often use knows best. And swat the jump into if you have thoughts on, I don't know, lots of numbers in the description and how helpful that is. Or maybe some of like the outlier, like the fact that yeah, Avengers is like very big and noticeable.

I mean, again, a flipped article is gonna say about like, understood fully well, like one year like, or like there's always been one down movie like Avengers or I don't write that on ones but if those are in the article, and like, you know that those are bigger and then you don't need to do it on text, but again, like you can get details like fine if he was gonna, it's gonna it's gonna add Scroll, scroll away, so don't don't say you don't want to sit through the old tech dude if you don't want to.

Yeah, it's a great point to

do. We maybe want to show some suggested alt text for this one. Let's do it.

All right, a chart showing the movies that make up the top four box office weekends in general the number one movie that takes up the lion's share of the sales for each weekend. But in the case of the Barban, hive and fire Bernheimer weekend, a team effort across Barbie and Oppenheimer and others brought the box office total over the 300 million mark and you can hear even in the in the most concise I guess version of that still pretty lengthy so you start to like see, I think too as you're writing it out, like okay, it's getting a lot of numbers in here too. And one one comment I was gonna make too is that like, even though we do try to keep it focused on the information, just because there is so much to communicate, and you are trying to do that pretty concisely. In some cases, you know, in this in this graphic if the if the graphics editor mentioned that this was very pink guy, because it's so unusual or noteworthy. I think I would just leave it in there. This is like the pinkish chart I've ever seen the New York Times published like maybe that is something to comment on in this case. Yeah, go for it

yeah, so in our examples, and of course, like you'd probably need to test across different graphics, but for us on the in the article, the screen reader would be able to access that. If this exact image was screenshotted, as we had here and share it on social media, all of that kind of gets burned into the image. And that's a case where we would make sure to like list out and say, a headline reads this. Some text underneath reads this and that's usually kind of the format we keep with that all texts in the middle for a social post. All right, okay, I'm gonna run through this piece quickly. But that's a little bit about, you know, the alt text that we are writing and the guidance and this portion is a little bit about how we got there, which has been alluded to it's really been a three year journey that started in 2020, with an article about the 30 year anniversary of the ADEA. And in this article, it was one of the first articles I think that we had published all texts throughout that package. And part of that was because so many of the journalists that were working on that story wanted to make sure this was very accessible, but it really was a test pilot for us and then showed that it could be done. And then in early 2021, the graphics department started to audit their graphics and start to make some affordances to be able to add alt text. And later that year, we started drafting guidance for photos and for illustrations. We started reaching out to groups like the American Council for the Blind to get some feedback on that. The field went live in our internal CMS and in 2022. And we started training there as well. We we also started doing some focus groups and some beta testing in 2022. And with that feedback, we kind of finalized our photo illustration and graphics guidance in late 2022. started training folks doing our own analysis of the Alteryx. That was written retraining. And that takes us to where we are now where we're looking into the fall and really this is a part of our workflow at this point. Okay, so we're just gonna dive a little bit into some of those key steps. I have to share some of what we've done there. But that's right. Okay, we have another slide where we just wanted to share I guess, like along the way, some of the really foundational questions that I think we had to answer to be able to get this across the finish line. And I think that this is all really good questions for anybody to answer if you're hoping to do this in your newsroom. With the first being why are you writing alt text? Who is responsible for writing and editing alt texts? The question that already came in came up as we've discussed this. What is your end goal? When do you aim to have alt texts across your site? Where will that text be written in your CMS? Where will it show up for readers? And how will you handle alt text insensitive situations and breaking news situations? You really do want to be sure that you have answers for these questions, because you will get a lot of questions from the newsroom. And not all of them will be easy to answer. So we just pulled some of like, I don't know, some of the challenging questions that I think the three of us have heard both at the times but also a swath of some that you've heard in your work at the American Council for the Blind. The first being that all text is difficult to write and maybe you can solo the photo, I have

a response. Okay, so, literally, this is a question that we would get in our newsroom. And my first question back to usually the editor who is saying this to me is like, Okay, walk me through your workflow. Like tell me how much time you're spending on alt text. I once heard that somebody was spending 15 minutes writing a piece of alt text for a photo and that was like an immediate thing for me where I was like, Okay, I know how to fix that because this should not for a more basic photo, or an illustration should not be taking more than one to two minutes max. To write all text. We're talking about one to two sentences. That is a visual description of what you can see when you look at the image. So that's, that's a big one. For me. For more complex images. I try to you know, would try to lean in more empathetically and sort of understand and also in cases maybe provide templates provide guidance provide anything that can make this process a bit simpler for the editor. While always grounding it in like this is the right thing for us to do. It is critical and it is required that we have alt text on everything so that our journalism can be as accessible as possible. But I want to throw it to Jamie because it's it's different for graphics Oh, no, no.

The dreaded your type Simon's. Yes. Yeah. And I think another point to make there too, is that if it's taking long, I think it's often been the case that we'll look at the alt text but folks are writing it. It's like incredibly detailed. And, you know, it's just so often the case that that the alt text itself could be very could be simplified, and that makes it easier to write as well. All right, we're busy and all text is more work. It is true that all text is more work, and it's true that we're busy. Both of those are true, but I think this is a question where it's really important to get back to the why why are we doing this? It really does take a fundamental shift and I think your perspective on what's important in this work. If you do believe that the information that you're providing to your audience and in this in this form is important. It's critical for people to understand than it should be critical for all people to understand and this should be an essential part of your work. This should be a part of our visual journalism, translating this work to anybody who can't access this visually.

And also like once once it's like part embedded in your company culture ever can naturally do so I mean, yeah, it's not gonna it may take longer we're dealing at first but we won't be doing it. So

yeah, yep, exactly. You get better the more often you do it, as we will see how many people are even using this? I think even if one person was using this, we would still think it's critical. This, this addresses and essential. This addresses a barrier that exists in our work right now. This is a barrier to access to our reporting. And so doing this work takes that barrier away and makes our work more accessible to a much wider audience of people. And again, we're going back to the why this is absolutely critical to include this swath I think this is one that you've heard before. We can't afford to add alt text or website

Oh yeah, we've heard a lot ACB if it's anything so again, like this technology already exists, we like that's our main like guideline and catchphrase, recall it. So it's already this. It's already part of like many platforms, so just it matter of finding it. And also, just even if not, like, already completed there. It's going it's not going to go off like the astronaut astronaut go mountain just add it so yeah, so there's so this and also just there are ways to like make it like not have thick brick brick bank. So

yeah. And I'll get to the to a little quicker because there's more of what I share. But can one person be responsible for all of the alt text? No, definitely not in a newsroom. There's so much visual reporting leads to the New York Times published every day. And like we said, this can be a very quick part of your process in a lot of cases and is it okay to publish a story and add alt text later? For us? The answer is no. And even this often comes up in like breaking news situations. And really, again, if this information we feel is essential for everybody to understand what's provided visually. That's true for readers who aren't able to see that as well. And for readers who use screen readers. So let me yes, you want to take from

here Yeah. So all these questions and the workflow questions that came up are reasons that it took time. But we tried to lean into it with empathy. We tried to go into it knowing that we would write down guidance and it would evolve and it would change we would have use cases that would test it and help us make other stronger decisions about again, how to prioritize accessibility in our newsroom. We began training once the field went live in our CMS. And just to explain a little bit what that is. We have a field where editors can write in what the caption is for the story. We did not have a field for alt text until 2022 As you saw on the timeline, which is when we were able to start regularly adding alt text to our photos and our illustrations. Jamie's processing graphics is different because a lot of that is hard coding and it's templates and it's using different tools. But we were really, really excited to get that field in our CMS and so just a note here, talk with your product people get aligned with them with like any accessibility thing. They're like great allies and advocates for the work. So we began some early training, we had some guidance, and we were kind of distributing it across our newsroom, and then we realized we missed a really important step, which is actually talking to people who use screen readers and seeing you know, what they thought what they need, what all text works for them whether our published alt text was working for them. So we did a focus group and we worked. The American Council of the Blind Spot. This organization helped us spread the word and helped us spread this ask to ask if people who use screen readers would be willing to test New York Times articles for us. So we sent a focus group of about 20 people six articles in six weeks and we asked them to give us feedback on these various points. I'm going to go through a little bit of what they told us. So this was a literal piece of all text that we sent to our our focus group participants during the test. It's really long. You can see that it's a pretty complicated visual we have people who are in wheelchairs. This was like a performing arts ensemble. That's people with disabilities. They gave us a lot of information in the alt text, including their race. A lot of like race and ethnicity information and the editor in this case chose to include it in the alt text. We got a lot of feedback that that was really confusing. People were like, Wait, why is it relevant? what race they are in this story. This story is like a criticism about a performing arts ensemble. And it's also really long and we don't know what at a 90 degree angle means. So a couple of tips here. This reconfirmed existing race and ethnicity guidance that we had already set in our newsroom which is only mention a race or ethnic component or other types of identity, when it is relevant to the story that goes for any identity so that like reconfirmed that it also reconfirmed our notion of like this actually needs to be way shorter. Here was another article that we tested you can see here we're talking about a bowl of Dan Dan noodles covered in slivered cucumber and crushed peanuts like so much description for this of food photography, which is awesome. You can see that that contrasts a little bit with the alt text. In this crowd shot which is an outdoor patio with plants and hanging lights, filled with casually dressed diners at simple metal tables. In the caption it mentions that this restaurant draws an enthusiastic and diverse crowd that confused one of our focus group testers. They were like, wait, what do you mean it's a diverse crowd? There's nothing about that in the alt text. I don't understand where that detail is coming from. So really think about what information you're putting in your alt text and what information you're putting in your caption. This was a literal piece of feedback we got from a focus group participant who told us the food was well described in the alt text. But all the humans may have well may as well have been the exact same person since no description was given. I'll let Jamie take this one on the graphic.

Yeah, totally. Um, we'll run through this quickly. It's a map of Europe that showing extreme forecast temperatures across Europe on Tuesday. That's the all text we were given. And so we're clearly describing what it is we're seeing what information is there but we're not saying anything about what that information was being communicated in the graphic. And that was not Law Center testers. So big tip here is make sure to to communicate the information. If you're using an information graphic, that's likely why we included it. They said the graphic clearly showed more detail about the temperatures in several countries, but the altex failed to communicate that in a clear way. And again, that reconfirmed a lot of the guidance that we were working on at the time there too.

So this first part is guidance that we already had in our newsroom that was reconfirmed by this testing. We also learned some things in illustrations we need to reduce the level of abstraction that's literally a thing we now tell our editors. We also tell them that in portraits, we need to describe more detail because the feedback we got was that screen reader users can understand it's a close up but like we need to communicate a little bit more about what's going on in that visual. The very last component here is so we went department by department in our newsroom we trained editors, my team did who work on photos and illustrations or work on stories that publish those types of images on how to write alt text. And then a few months after we trained the departments, we actually did an analysis of their alt text and we did it like in a point in time. So we took all of the old texts that one department published in a month, like using a data scraper, and we looked at it to understand like how many stories were getting all texts in many cases, a lot of them were so that was awesome. And also like what was the quality of the alt text? And here's a few key points that we found which I'm just going to share really quickly in case they might come up in your newsroom. A big one name the person this is what we talked about in our Margot Robbie and Ryan Gosling example. But you can see here that the alt text says an older man in dark glasses and a blue suit standing at a lectern. But that man is President Biden. And we need to say that because first it's very ambiguous who he is. If we call him an older man. Second, we probably shouldn't be referring to people as older men or people in our alt text we can give a clear description. The second is give some detail. So instead of just saying the Dumbarton rail bridge, we need to give some more detail than that. A view of a part of the Dumbarton rail bridge as seen from train tracks which run over a body of water. The last is keeping the context to the caption. This is a big one, Jamie, I know you also see this come up in graphics. Definitely.

It's like very often, even though there's already so much information in a graphic. I think there's a temptation to try to explain some of the contexts that might not be there because it feels natural to try to do that when you're giving some some visual description. Avoid that, again, we really want to be as concise as possible. It's confusing and it's misleading. It's not information that we're getting from the from the image or from the graphic as well. And that's an important takeaway to

absolutely, um, I can just go through this real quickly, Jamie, um, some steps to building a culture of accessibility in your newsroom. It all starts with the fact that this is the right thing to do. We are prioritizing making this an inclusive experience for people with disabilities and people who use screen readers and everything starts there and then builds from it. We really recommend tiny experiments. To prove in your newsroom that this is possible. And not only is it the right thing to do, we can do it and everything kind of comes from there. Please use our guidance. This is on our session page. We also have a QR code and a bitly link bit.li/all-tech. Stash o n a 23 I'll give everybody a minute to get that down. I also am going to show a few examples of how to put this into your content management systems. Since we started five minutes late. I want to leave the remaining time for questions. We also have a table talk that starts at four. So please ask your questions. here or come join us and hang out with us at the table talk at four. I want to pause now and take any questions. Yes.

You're with us I've worked on my audience for the workflow trying to communicate with audiences

and their social, yeah, great question. So here are literally some ways to do this on social platforms for anyone in the room who hasn't experimented with that yet. But for us, it's like it's all alignment. So it is you know, making sure that story editors are putting all text on the photos and the illustrations and the graphics so that that naturally translates to social. So it's also at the point where social can the audience team can then cribbed the alt text that is already published on the image. And as Jamie mentioned, like add to it. So you know we're taking the publish all text, we're adding additional context that it needs for social we're spelling out the headline we're putting in the source. So yes, they're like two different platforms. But I really recommend like getting everybody on the same page because it's only going to help everybody and I think from an audience standpoint, if the story editors are already writing it that makes your job like that much easier when also it's like the editors who are the people with the context and the nuances of the story, who are taking the stab at the alt text. That's the process that like we kind of have, but I know that that's not possible immediately in all newsrooms. So the big other takeaway is like some alt text is better than no alt text. So I've heard of like places where like it's just the audience teams who are writing the alt text that is also great. And hopefully these tips can help encourage the conversation in like the wider newsroom to

fight back on the restaurant, example. And the caption says something about a diverse crowd, the Old Testament with criticism was because there's a new idea being presented as a caption that wasn't represented in the old texts. Is that

correct? That's correct.

So in that, in that case, then would you go and describe the people sitting at each of these tables? And in doing that, are you assuming

their race? I actually in this case, would probably take out the word diverse from the caption and I would probably maybe leave it at draws an enthusiastic crowd, because the diversity component is confusing, and it just feels a bit like that could be the rewrite that would sort of fix the issue there. Does that help

to a two sided reader as well right like so. So often, I see that a graphics you like so often, if you're having a hard time writing to that to that point, it's it's probably a challenge for a visual reader to make that connection as well. And same is true of graphics. If it's really hard to write a big takeaway, maybe that takeaway is not obvious to a sighted reader as well. Yeah.

I also didn't see any very good looking at asset so

there are two letters that have come up a fair amount at this conference. They are a I side conversation, but we're just in lesson was, Oh, I'll text the great news for AI because you can have an AI within the picture next night imagining this has come up with your organization. Super curious where

Yeah, so for us, like texting help helpful, but also like, it shouldn't be your main tool in anything. It should be a good supplement. So like, you know, you can have alt texts like right, and Alteryx AI, right, I'll fix for you, but you have to install check it off, make sure it's accurate and that like I was talking at lunch with Jamie about the big if not called Be My Eyes that it's for blind people to connect with. Sighted volunteers to help out help describe things that can help interpret visual emission. So like you can have the call and be a governor call and ask them Hey, okay because because my shirt is different this shirt go with might go with my jacket. So like that. They have never tie with AI like having ai do bad work. And the I think Jamie mentioned that she did use AI to try and try and describe the Barbie Barbie graphic that Robin Hood graphic and it didn't like capture all the details in the graphic. So it got stuff wrong. So like it's still it's a great tool, but also just not one thing. You can't you have to you can't like rely on for all of your Eltechs so

yeah, and I would just would love to add to that if I can that like I'm so grateful that we've started this work, maybe at the precipice of this conversation because you know, even with these tools becoming available, and maybe we will find a helpful way to like use AI to augment our process. So much of this work, I think of as remediation work. This is work that should have been done that shouldn't have been done across all newsrooms a long time ago. And so so much of the work of figuring this out and adding this into our workflow is really taking a look at I don't know how, who we're writing our stories for who's in our audience and how inclusive we're being. And I'm so grateful that we're making that pivot and the Baltics is there in our process now for us to pause and reflect on that. This is one piece of accessibility work gets doesn't solve everything. And, you know, I hope we all continue to do that along the way. Yeah. Oh, sorry. Well, good

question. Was there anything that's been the right thing to do? Or was there anything that has happened? I mean, like universities, lawsuit, there won't ever be universities, PowerPoints, or anything.

The other thing we would bring up in our trainings is this is required from the Americans with Disabilities Act, but we focused it all in this priority for making it an accessible and inclusive experience for people who use screen readers. Oh,

also on that, it's been about both ADA. Ed has like effective, effective communication for the title so with that means it's like it people who people disabilities are entitled to that back to communication and if that looks different than need to cover that so yeah, if like if you if you're communicating that doesn't translate to someone disability, there has to be a kind of firm may or may or mediate, and so

yes. Legally, we're not required to all of this work. The process the focus, we pride ourselves on I'm so tired of the exercise and the data into into considerations. They're tied to the price itself and the quality of our graphics. It's not just the fact that it's an enjoyable experience.

So everybody should have

that moment in the experience, whether they can see the graphics or not. So it's, it's not only the right thing to do, but it's doing at the level that you would expect from the right so it's not just Caribbean best practices. You want to pass the legal check. Test, instead of my common with our journalistic standard covered there, and that's why all the testing that's why it took so long to get to a point where this is part of a workflow because we wanted to bring everybody in understanding this is not just something we're required to do, but we want to understand that everybody in our audience expects and all of our readers whether they're scientists are really artists of their level of science experience.

And I saw a question over here, sir. Conversation,

sensibilities, it goes beyond this two stories a consequence of New York Times so interactive, and it's just like to remind you that then social media like you know, we have some options, but we don't promote Instagram or something because that's something that

you will we we do have it on Instagram. Instagram Stories, no, not on stories. But we have brought it up with meta. Just saying, but this is how you do it on Instagram. But I know you had a no, it's this is like right when you are about to hit share, you go to Advanced Settings and it's under right alt text. That's how you get to that oh, yeah, right. It's not in third party tools. It is in social flow. If you Yes, yes.

Yeah. There's lots of new social media.

We can talk more at the table talk if you want to. Yeah. I think actually, because we're a bit over what happened but we're here so come up if you if you have a question and just wanted to say thank you, everybody for joining us today. So chatty Maddie sensibility.

You