ABCDEFGHIJKLMNOPQRSTUVWXYZ
1
Table description: This table presents feedback from arXiv users and accessibility experts on their experience accessing research. This header row lists scientist's five main goals when interacting with research. The second rows describes positive experiences that support access to research at each step. The third row displays a graphic of a bar chart for each step. The fourth and last row describes negative experiences that are barriers to access. Find researchRead researchParticipate in scholarshipPrepare my documentSubmit my paper
2
Promotes accessibilityFinding research on arXiv was noted as "not a big problem" and "OK" by researchers using assistive technology. One benefit noted is that instead of having to search through multiple journals, arXiv covers many fields which makes finding papers more efficient. "It is a one stop shop for everything I am interested in." In addition, researchers do not need to log in, the layout is simple and fast loading, and the website works with screen readers.
Reading research is a largely inaccessible experience, but some positives were pointed out as well: arXiv overcomes financial barriers; the arXiv version sometimes has more content than the published version; and TeX source is more accessible than PDF.
Over 30 years ago arXiv sparked a revolution in scientific publishing by allowing researchers to share their work quickly and directly with each other and without financial or institutional barriers. As one researcher told us: "I see arXiv as a continuous discussion within the community that never stops.” This is critical work that arXiv must continue and expand upon until everyone can participate, regardless of disability.		Preparing documents is an exciting yet stressful time for researchers. Experiences that promote accessibility when preparing a document include when a conference or journal mandates it, or when the technologies they already rely on have high native accessibility or easy to use tools.
The arXiv submission interface is familiar and comfortable. It is also free, and there is no deadline: it is open when your research is ready. All these are aspects of access that are important. Thinking about future improvements, one researcher said that "arXiv has a closeness to the practitioner that is exciting" for accessibility efforts.
3
Image description: This row display a positive/negative bar chart for each step. The underlying data is available in the second sheet of this document, titled "data". Each chart displays positive feedback above the mid line and negative feedback below it. The height of each bar represents the impact score of each piece of feedback.

Of note is that the step "read research" received the highest amount of feedback by far, and it was overwhelmingly negative in regards to accessibility. Ensuring the reasearch that arXiv hosts is accessible to all is the next logical step for Open Science.
Find Research received 11 total observations, 38% of them negative.
Read Papers received 57 total observations, 83% of them negative.
Participate in scholarship received 17 total observations, 68% of them negative.
Prepare my document received 27 total observations, 89% of them negative.
Submit my paper received 14 total observations, 57% of them negative.
4
Barriers to accessibilityDespite the positives of arXiv's website, researchers do face barriers to finding the content they need. For example, searching novel or complex terms with voice command is challenging because spelling errors are difficult to detect via audio playback. arXiv's daily emails are particularly inaccessible due to their length, low usability, and non-specificity of subject matter; in some fields the output is so vast that the daily emails are not useful. Most importantly, arXiv, as in academic culture at large, prioritizes PDF over more accessible formats. An HTML option would greatly benefit users of assistive technology. The majority of feedback on the accessibility of reading research is negative. As one blind scientist told us "progress is not moving anywhere on accessible papers." Most research papers still suffer from unparseable figures, math, and graphs, and are not formatted in a way that allows assistive technology to navigate with efficiency. One blind researcher shared how they were forced to change working groups because, due to inaccessible PDFs, their read rate was below the groups' target metrics. The problem was described like this: "Creating accessible documents is a solved problem technically but is difficult to achieve in practice due to our choice to publish documents as PDFs." Most accessibility issues would be mitigated or resolved by providing well formatted HTML.

Full participation in science requires more than free and open access. As a NASA communicator told us "There are things we can do to welcome people, but if they can’t participate fully it's all just nice talk." Full participation demands fully accessible research output as well as new cultural norms.

Participation goes beyond digital access: One researcher shared how they hide their disability in professional spaces: "Mathematicians will offer to help with a staircase, but wont engage me in a conversation. So I tend to hide [my blindness] in professional settings." Another explained that they have almost been forced out of their field on multiple occasions but, due to the pioneering accessibility work of just a few individuals, the necessary tools became available just in time. 		The consensus we heard from experts was summed up in this way: "Even accessibility researchers are bad at making accessible papers. It’s really complicated." Preparing an accessible PDF document requires special knowledge and proprietary tools that are cumbersome and sometimes prohibitively expensive.

That said, we also heard that authors have a very important role to play in creating accessible research. But they need tools and guidance. “I think one of the biggest problems that we have is the paucity of tools that make it easy to make HTML.” Many people we spoke to envision a future where writing clean semantic markup or sharing underlying data is the norm.		Barriers to accessibility include the difficulty of troubleshooting TeX, which robs authors of the time and brainpower to think about other aspects of the submission process; A lack of structural accountability, such as prompts or alerts that certain content is not accessible; LaTeX packages intended to improve accessibility but that instead cause technical issues; A lack of accessibility guidance and tools at the point of submission that could allow authors to improve their document; And lastly, the continued dominance of the PDF format adds layers of complexity and blocks the development of better solutions.

The message from our diverse participants was clear: What arXiv can do is generate accessible HTML in submission pipeline.
5
End of worksheet.
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100