Three big things since the mid-August update: a new toolbar, the results page has been redesigned around one list at three densities, and looking up or re-formatting a long reference list is dramatically faster. Plus a pile of smaller things — including a few you asked for.
Much simplified top toolbar
The top toolbar was very crowded, and much of what was there was little-used. Now, it’s pretty sparse.
A Session menu on the left saves, loads, or clears the current references. A formatting menu on the right shows the current style and opens to reveal the style picker plus the font and size controls. In the middle is the List / Diff / Cards switch, discussed below. If you’ve loaded a file, you’ll also see its name and a Reload button to process it again (presumably after you’ve made changes!)
One list, three ways to read it: List / Diff / Cards
The results used to include cards with lots of useful information, like this:
This shows the original reference, the suggested reference, the differences between them, what changed, and where the information came from. It’s a lot to take in, especially when your reference doesn’t need changing.
There used to be a separate set of cards that showed just the diffs, but the new interface has a single list that can change shape. You can also see just the diffs:
Now you can quickly see that the title got changed and that the journal volume, number, page numbers, and DOI were added. For a long list, though, that’s still a lot of information that you don’t care about, so there is a new compact view called list:
Here we see both the reference from the past two screen shots and also one that needs no attention. The first one has an indicator that we added a DOI; both have a green 
indicating that the information has been verified.
Clicking a collapsed list item opens it up. Each card has a 
to let you choose what view to switch to. The same selector at the top of the page changes all the cards at once.
To recap:
- List — one line per reference, with every badge the full card would show. Scan the whole list in one screen; click a row to expand just that one.
- Diff — what you cited next to what we found, differences highlighted. Click a row to choose that version.
- Cards — the full card, as before.
The cards themselves got quieter. A verified reference shows the verdict and little else; the Retry/Report buttons, source icons, and metadata only appear when something needs attention or when you open the details. The upshot is that the references that need you stand out, because the ones that don’t have stopped shouting.
Approvals now mean something. When RefRunner auto-picks a version, the button says · auto until a person signs off on it — and you can approve a kept original too, not just a replacement. The score sheet has a new unapproved tally so you can filter to “nobody has looked at this yet”. It’s kept separate from “needs review” on purpose.
Much better performance on long lists
A 210-reference list used to take minutes to reformat when you switched styles. It’s now about ten seconds — and switching between styles you’ve already used is nearly instant.
What changed under the hood: we were building a fresh citation engine for every reference (about 700 ms each); now there’s one engine per style (about 10 ms per reference). Each card is also formatted once, after its enrichment finishes, instead of once per intermediate update, and off-screen cards skip layout entirely. Lookups keep running when the tab is in the background, and the tab title shows progress and tells you when the run finishes.
Chrome runs the foregrounded tab a bit faster than ones in the background, but if you have a long list and get tired of just watching it do its work, you now have an indicator when it’s done!
While it’s working, it looks like this:
and when it’s done, you can see it clearly:
If you have a zillion tabs open so that all you see is the little icon, you can still tell by the orange dot or green 
on the icon. Sorry to use so much space showing this tiny detail, but I was very excited! (I am also excited that it processed those 32 references in under a second each! 
)
Smaller things
- Import RIS and EndNote (.enw) exports, alongside
.bib. The drop zone now says so. If you use a legacy reference manager or get a list from Google Scholar, you can now open it with RefRunner. - Notes on a reference. End an entry with
%% your noteand it’s kept as a note on the card and hidden from every output — the formatted list, the export, the copy. Handy for “check this with the author” reminders. I use this for my sample documents to explain why the reference was handled like it was. - Reload file The new simplified toolbar now has a RELOAD button to process your file again after you’ve made updates suggested by RefRunner. A keyboard shortcut (
t l) will save you a click!
- Improved phone support. Logs show that some people are using RefRunner on their phones, so we made a pass at improving the mobile experience. For example, mobile devices now show the List view by default—tap a card to cycle List → Diff → Card.
- Scopus-delisted journals get flagged — but only for work published after the delisting.
- ERIC now verifies and formats education references that Crossref doesn’t know, with a link to the ERIC record.
- Multi-word and particle surnames (van der, de la…) are better supported in Vancouver and Chicago; diacritics that publishers strip from author names are kept from your citation.
As always: if RefRunner gets something wrong, hit the “RefRunner got this wrong?” button on the card. Several of these started that way. You can also send a message to @pfaffman (or email me). If you suggest something, there’s a good chance it’ll get implemented or fixed.