[seqfan] Re: OEIS Frontend Redesign

[Arthur O'Dwyer]

On Sat, Jan 27, 2024 at 8:54 AM Alexander Craggs <oeis at femto.dev> wrote:

> Hey,
>
> I'm a big fan of the OEIS website (having used it for just over a decade
> now), especially its ability to understand sequences covering a
> surprisingly wide range of topics.  My friend and I stumbled across the
> 'Suggestions for OEIS from OEIS 50 Workshop <
> http://oeis.org/wiki/Suggestions_for_OEIS_from_OEIS_50_Workshop#General_remarks_about_making_changes_to_the_OEIS>'
> page on the Wiki and saw a 2014 post about updating the site to HTML5 and
> improving mobile accessibility.  The accessibility and user experience
> improvements suggested don't seem to have come to light, the design having
> not significantly changed since 2012 [1] <
> https://web.archive.org/web/20101129053330/http://oeis.org/> [2] <
> https://web.archive.org/web/20130309111854/http://oeis.org/search?q=1%2C2%2C3%2C6%2C11%2C23%2C47%2C106%2C235&language=english&go=Search
> >.
>
> We wanted to try to change this, so we took a couple of days off work in
> order to redesign it.  We focused on the following:
>
> - Improving entry-level use.  Instead of requiring searchers to know about
> 'keyword:' and 'author:' tags we now include buttons to help out, whilst
> still allowing the full complexity for advanced users.
> - Mobile-responsive design, ensuring text is legible and buttons clickable
> on all screen sizes.
> - Keeping the site fast and small.  Caching allows us to save 88% of data
> over 100 sequence searches (3MB new vs 25MB original).  We're also faster
> on first load, the query '1, 2, 5' being 33% smaller (600KB new vs 900KB
> original).  The page load delays are primarily caused by the OEIS server
> responding slowly, which aren't easily fixed by us.
>
> The initial version is live here:
>
>     https://intdb.io
>
> It's an open source project, so you can also see the source code here on
> GitHub, <https://github.com/popey456963/intdb> we accept contributions
> from anyone!  We've done this in a 'hackathon' format over two days, so
> there are numerous bugs listed here <
> https://github.com/popey456963/intdb/issues>.
>
> We know it's missing a large number of key features like accessibility,
> i18n and various links (especially my favourite 'listen' feature), but
> wanted to put it out there to see what people thought and whether there
> were any suggestions on ways we could improve things.
>

IMO all the technical details you've provided (smaller, faster, etc) sound
great.
But personally, I like the "Web 1.0" style of interface much better than
the rounded-corners overly-CSS'ed Web 2.0 style you've gone with. Even if
you keep everything else 2.0-ified, I suggest that you might get better
reception from the old fogeys (and I'm not even 40 yet!) if you *merely
switched your color scheme* to #ffffff + #0000ff + #fffcc.
For example, one easy psychological improvement would be to switch your
purple color (which I can't figure what color it is because I can't figure
out where your HTML lives, even with "Inspect" in the browser!) to #0000ff.

I also suggest that the high visibility of "Webcam" right next to "Privacy
and Legal" may be a trigger. ;) You should move the administrivia into the
footer and separate it from the "features" like Webcam (which IIUC is
really just a pre-existing cute name for what StackOverflow calls "Hot" and
Wikipedia calls "Recent Changes").
I recommend
- https://www.webdesignmuseum.org/gallery/google-2002
- https://www.webdesignmuseum.org/gallery/google-2005
as the right inspirations.

As an only occasional OEIS user, I didn't know about all the
shortcut/keyword stuff. Pushing "Keyword" into the new user's line-of-sight
isn't great, unless it does what the newbie would think it does. I clicked
"Keyword" and typed "polygons": no hits
<https://www.intdb.io/search?q=keyword:polygons&sort=relevance>. Just like
the existing OEIS search, if I want to find results about polygons, I need
to enter "polygons" with *no* prefix. I'm not sure what "Keyword" does,
exactly. You could use some of that white space to the right of "Author",
"Keyword", "ID" to explain briefly what each of them means.
("author:Neil Sloane" doesn't do what I'd expect, either.)
For example,
  Keyword — e.g. nonn, more, 2d, ... [Maybe?]
  Author — e.g. _N.J.A.Sloane_ [Maybe?]
  ID — go directly to a sequence, e.g. A051493
Basically, if you're pushing something to the front for the benefit of the
newbies, it ought to make their lives easier, not harder.

But! A simpler and less newbie-frustrating approach would be to kill the
dropdown and instead put some plain text right below the search box that
says something like "*Click here* for advanced search terms id: author:
keyword: [etc]", and then the link would just go to a page of documentation
(or unhide a div of documentation, if you want to be all Web 2.0 about it).

Getting back to the visual design:
- When you're trying to "steal market share" from a "competitor" (not in a
bad way, mind you!) you want to reduce the psychological friction of
switching, as far as possible. That's the main reason I recommend matching
the color scheme and "Web 1.0" styling and so on: less friction for my
brain.
- But also, there's nothing *wrong* with the existing site's styling of *search
results* AFAIC. The main thing I think of as "wrong" is the attributes on
the text boxes to *edit* a sequence (last time I checked, which I admit was
a while ago). Your site right now doesn't provide editing; I assume that's
because it doesn't provide login; I assume that's because it's
technologically impossible for it to log people in nor check their login
status without actually *being* oeis.org.
- Have you thought about how to at least mock that up? (Like, your site
could provide a "Pseudo-log in" button in the upper right corner where the
"Log in" button would normally appear; and then switch into a "logged in
mode" where the "Pseudo-edit" button appears; and so on. Even if you
couldn't submit valid updates to OEIS, that would satisfy the goal of
showing the mockup to people.)


P.S. This is the first time posting to a mailing list, and I've really
> struggled.  Hopefully this email goes through, unlike the last couple of
> attempts!
>

It went through! :)

–Arthur