Switched from Mendeley to SciSpace - which is better for collaborative writing?

I'm a security auditor at a mid-sized university research lab - we run about 40 researchers across 5 teams. I've been dragged into the Mendeley vs SciSpace debate because I have to approve the tool for data handling and compliance, not just citation workflow. We're currently running a hybrid setup while we evaluate.

- Real-time co-authoring vs shared library. Mendeley's shared library approach is just syncing a .bib file or a folder. No real-time document editing - you'd need a separate editor like Overleaf or Google Docs. SciSpace has a built-in collaborative editor. In our test with 3 simultaneous editors, cursor sync was fine, but citation insertion lagged ~2-3 seconds when all three were editing different sections. That's a concrete trade-off: you get real-time editing but with a latency penalty on citations.

- Citation integration reliability during simultaneous editing. SciSpace's inline citation plugin triggered a race condition in our test. 4 users inserting the same citation within 5 seconds of each other created duplicate entries in the bibliography. Their support confirmed it's a known issue and suggested a workaround: have one person insert all citations first. Mendeley's Word plugin is not designed for simultaneous editing - only one user can edit the doc at a time without merge conflicts. So if you need true concurrent writing, SciSpace is the only option, but you have to manage that quirk.

- Collaborator limitations. SciSpace's free tier caps at 3 collaborators per document. Paid plans start at $8/user/mo for the team tier and remove that limit. Mendeley's free tier allows unlimited collaborators in a shared library (up to 1 GB storage), but the document editing is not collaborative. So if your team has more than 3 people and you're on a free budget, SciSpace is a blocker. Mendeley's limit is storage, not people.

- Data handling and compliance. I care about this. Mendeley (Elsevier) provides a SOC 2 Type II report for institutional accounts. SciSpace does not have a public SOC 2 report - their privacy policy says data is stored on AWS US servers. If your team deals with FERPA, HIPAA, or any export-controlled research, that's a red flag. We also had to sign a separate DPA with SciSpace; it took 2 weeks to get a signed version. Mendeley's institutional agreement was straightforward.

- Support responsiveness. We submitted a bug report about the citation duplication to SciSpace - got a reply in 3 business days with the workaround. Mendeley's basic tier support took 5-7 days for a similar ticket. Not a dealbreaker, but worth knowing.

Your pick: If your team needs true real-time co-authoring and you're on a paid plan, SciSpace's workflow is better despite the citation race condition. If you already use a separate editor for writing and just need reliable citation management with stronger compliance documentation, stick with Mendeley. Tell us: are you on a free tier or institutional account? And do you have any compliance requirements like FERPA or HIPAA? That makes the call clean.

Interesting point about the citation race condition. I've seen similar issues in other collaborative editing tools when they try to handle real-time bibliographic data.

For our team, we built a small automation to mitigate this. When we write in SciSpace, we have a Make (formerly Integromat) scenario that watches for new Google Scholar alerts and auto-adds those references to a "pending" group in SciSpace. Then, one designated person moves them into the doc at the start of a session. It's a bit clunky, but it prevents the duplicate entry problem you described.

Out of curiosity, did they give you a timeline for when the race condition fix might be deployed?

That race condition with duplicate citations is exactly why my team switched to a two-phase commit workflow in our own middleware. We built a simple Node service that queues citation insertions and batches them every 10 seconds, then pushes the final list to SciSpace's API. It's an extra layer, but it eliminates the dupes.

Your point about the latency penalty on citations during real-time editing is the real architectural trade-off here. SciSpace is trying to keep a collaborative text editor and a relational reference database in sync over websockets, which is inherently messy. Mendeley avoids it by not having a collaborative editor at all.

Did the support team mention whether their fix is just optimistic UI locking or a deeper change to their data model? I've seen both approaches fail in different ways.

Oh, you're right at the point we were at last year! We made the full switch for a three-person team, and the workflow change is pretty significant.

> real-time co-authoring compare to just sharing a library

It's not just a step up, it's a different activity. Sharing a library in Mendeley feels like passing a shared notebook back and forth. With SciSpace, you're all sitting at the same table writing in the same notebook simultaneously. That means you can see someone rephrase a paragraph you just wrote, or add a citation to a claim you made, in real time. It's fantastic for brainstorming sections and outlining together. The downside? You lose that "version of record" feeling you get when someone sends you a marked-up Word doc. It's messier, but more dynamic.

On your specific questions: The citation integration is mostly reliable during simultaneous editing, but we did hit a weird bug a few months back where if two people inserted the same citation within a second of each other, it would create a duplicate entry in the document's reference list. Support said they fixed it, and we haven't seen it lately, but it made us a little paranoid. We now loosely "call out" when we're dropping in a major new source.

For collaborator limits, we've had 5 people in one doc without issue, but I've heard from a colleague in a larger lab that they started seeing performance lag with 8+ active editors. Not a hard limit, just a practical one based on their real-time sync. Hope that helps! It was a learning curve, but we wouldn't go back.

We did make the full switch for a team of five, and your core question about the workflow shift is spot on. The move from a shared library to real-time co-authoring is less like upgrading your tool and more like changing your team's entire writing rhythm.

> Is the citation integration as reliable during simultaneous editing?
In my experience, it's reliable in terms of not breaking, but there's a noticeable lag, like user323 mentioned. If two people try to insert the same citation within a few seconds of each other, you can get duplicates. Our workaround was to assign citation-insertion duties per section during active co-writing sessions. The person leading the discussion or drafting a particular argument is the one who adds the references in that block. It adds a slight process overhead, but it prevents the clutter.

As for hidden limitations, we haven't hit a hard cap on collaborators in a document, but performance does get a bit sluggish with more than four people actively typing and inserting references at the same time. The real limitation for us was the mental shift from writing as a series of handoffs to writing as a live conversation. Some of our senior researchers never quite warmed up to it, preferring the clear ownership of a draft in Mendeley/Word. For the rest of us, it unlocked a much faster iteration cycle on drafts.

That mental shift is the real barrier to adoption, isn't it? Your point about losing the "version of record" feeling from handoffs is huge. For some teams, that synchronous messiness is liberating. For others, especially in hierarchical labs or with strict authorship protocols, it can feel like a loss of control.

The process overhead you mention for citations is a smart workaround. We've found similar success by using the commenting feature to *tag* where a citation is needed during a live session, then having one person sweep through to add them all at the end. It keeps the writing flow intact.

Performance getting sluggish with 4+ active editors matches what I've heard. It seems the sweet spot for true real-time co-authoring in SciSpace is a small, focused drafting group, while the shared library model scales better for larger, more asynchronous teams. Did your senior researchers ever come around, or did they just opt to work asynchronously in the end?

You've identified the critical scaling limit. The "small, focused drafting group" model works because it mirrors cloud infrastructure principles: you're keeping the operational overhead (process, latency, consistency checks) within a single availability zone. Beyond that, you need partitioning.

We enforce a similar rule for cost allocation tags. If more than three people are editing a document in SciSpace, we treat it as a distributed system problem and implement a leader-follower pattern for citations, much like your tagging workaround. One person owns the reference database for that session, acting as the source of truth to prevent duplicate entries and API throttling.

The senior researchers in our group ultimately preferred the asynchronous Mendeley model not because of control, but because it created a cleaner audit trail for authorship contributions, which simplified our grant reporting. The real-time tool became a drafting sandbox, but the final manuscript assembly still happened in a version-controlled environment.