I really liked the ability to comment on individual cells in old Observable, and I think the new notebook comments are not as useful.
My main use case is for education, where I have students fork an “assignment” notebook with multiple questions in it - each in a separate cell (often with other cells demonstrating concepts, etc… along the way). The students then answer the questions by adding cells with their answers (which might be text, code, etc…). They then shared their notebook back with me when they are done. I then marked the notebook by commenting on their answer to each question.
Yes, I can just type all of the feedback in the comments at the bottom of the page. But this is clunky for marking, because would I have to constantly jump back and forth between the answer and my comment, and it is also clunky for the student to read, because they will need to jump back and forth as well.
More generally, I find that comments on a document are most useful when they are close to the specific content they are commenting on. For example, the editing comments in MS Word, or the discussion comments in Perusall.
Please bring back contextual comments in Observable.
As another educator, I agree. One of the aspects that characterise notebooks is their emphasis on narrative, both textual and visual. Being able to comment on the subtleties of the development of that narrative is useful. Equally, commenting on specific code cells that are reactively linked can provide useful feedback on how we assemble elements of code into a whole.
We’re considering two primary use cases for comments.
The first is editorial review. This favors close collaboration in a high-trust environment, often with authors and commenters working together towards a shared artifact such as a report or analysis or dashboard. The commenter here is typically serving as a reviewer, editor, stakeholder, or teacher. Hence attaching comments to specific pieces of content is useful for precision, locality, and rapid iteration. This style of comments is typical of collaborative document editing software such as Google Docs (and MS Word as you mentioned).
The second is public discourse. This instead favors loose collaboration across trust boundaries, say for unaffiliated readers responding to a published article and having follow-on conversations amongst themselves and/or with authors. For this use case, it is desirable to separate public comments from the author’s work so that readers may experience the work uninterrupted and without distraction — before choosing to read responses, or not. And for commenters too, it’s good to encourage reading the whole work (“RTFA”) before responding. This style of comments is typical of blogging and publishing platforms such as Medium, Substack, and most journalism outlets.
I don’t think we can satisfy both of these use cases with a single design, and we’re currently prioritizing the second use case (community over teams).
Our old design favored the editorial use case, so it’s understandable that you miss it in an educational context. In Old Observable, comments were only visible by default to authors and so the vast majority of readers never saw comments. In New Observable, all readers see comments, which makes them much better for public discourse but necessarily means they live below the notebook so they don’t intrude. (You can imagine how choatic it would be if we allowed any reader to insert comments anywhere on a notebook!)
I’d like to support inline comments in the future for teams (our primary paid tier), because that is most appropriate given that teams are for close collaboration with high trust. But in the short term we’re favoring public comments since those are free for everyone. Not ideal, but if you really want localized comments now and you’re working in a team, consider editing and/or marking up content in the notebook directly. We can work on adding inline comments for teams in the future but we have a lot planned already this quarter so I can’t say how quickly it will be available.
Thanks as always for the thoughtful feedback!
Thank you for taking the time and effort to engage so meaningfully with your users. I really appreciate it!
I understand the distinction you’ve made here (editorial review vs public discourse), and I can see why you would prioritize the later.
I don’t think we can satisfy both of these use cases with a single design
In the spirit of thinking about alternatives…
What if the comments were per cell, but moved over into a right-hand sidebar that could be toggled open or closed (like the AI agent sidebar)? Perhaps they would default to closed for the public and open for the team. In addition, the author could decide whether comments can be made by only the team or also the community.
I get that comments on blogs and articles are typically “global” and “under” the content, but a more granular approach could be a differentiator and encourage more specific and substantive discourse, instead of the “quick take” typical of so many online comments. Check out Hypothesis for a (rather elaborate and education focused) example of this approach.
We just shipped a new comments page (release notes) which reinforces our emphasis on notebook-level comments as it lets you browse recent comments across all your notebooks. Our primary goal here is to make it easier for authors to read and respond to public comments (as prompt responses are a big part of what makes public discussion work) but it’s also useful within a team context to have more visibility into ongoing discussions without needing to check individual notebooks.
In this new cross-notebook comments feed, I expect cell-level comments would be less effective as you wouldn’t see the cell the comment is attached to, making the comment harder to interpret. Notebook-level comments by nature encourage commenters to include a bit more context in their comments, and hence may make those comments easier to read out of context. We’ll see.
I still think inline comments would be useful for review, though, and I’d like to get to that eventually.