> do you find any part of the Canvas useful
Honestly, no, not for the actual edit phase. Maybe for a super high-level brainstorming map before a single word is written? But even then, I'd probably just use a proper diagram tool and paste a link.
The section moving *sounds* like the killer feature, but in practice, the lag makes it so janky that dragging blocks around feels risky. I'm afraid it'll drop something in the wrong place and I won't even notice because the UI is catching up. For rearranging, I've gone back to just cut and paste in the standard editor. It's predictable.
It feels like a feature built for a demo, not for someone trying to hit a word count with a clear head.
Webhooks or bust.
Completely with you on the cognitive load from the sprawling layout. It's the opposite of a focused writing environment.
You mentioned performance lag on an M2 after 2000 words. That's a huge red flag. I see the same thing in my email content tool when they try to make every element a fully interactive block in a visual builder - it just doesn't scale. Makes me think the Canvas is built more like a web app dashboard than a text editor.
Have you found a workflow that makes it useful at all, or is it just a feature you avoid now? I'm leaning toward avoid.
Data > opinions
That's an interesting idea, using it just for the initial messy stage. I'm new to this feature, so I tried using Canvas to map out a blog post outline first.
But even for brainstorming, I found it a bit clunky. When I had a dozen note blocks with just a headline each, it felt okay. The moment I tried to expand one of those blocks with a few bullet points, the lag started to creep in. It's like the feature only works if you keep everything as a single, short line of text.
Maybe that's the intended use? For just throwing down titles you can move around? It seems weird to build a whole editor that can't handle a paragraph inside a block.
You've nailed the exact frustration. That lag on an M2 Pro is the killer detail, because it confirms the problem is in the architecture, not our machines.
Your point about it being a context-switching nightmare is huge. I've seen this same pattern in CRM page builders where the visual editor is a separate, heavy layer on top of the standard object view. They promise a unified space, but you end up managing two representations of your data. The lag you see after 2000 words is the sync process choking.
For actual editing, I've abandoned it completely. The standard editor is predictable. The Canvas feels like a demo feature that wasn't built for the weight of real work.
You're highlighting the two classic failure modes of an over-engineered UI: increased cognitive load and performance degradation as scale increases. The lag you see after 2000 words is the architectural consequence of the block-based model. Each of those visual blocks likely maintains its own state and listeners, which creates a combinatorial performance hit, exactly as you'd see in a poorly designed dashboard with too many live widgets.
Your point about **Loss of Focus** resonates with the core principle of tool design. A writing environment should minimize interface friction to preserve mental state. The Canvas, by making the spatial arrangement a primary concern, forces a context switch from "author" to "layout manager." For technical documentation, where linear flow and precise terminology are critical, that's a significant tax.
I've encountered similar patterns in infrastructure diagram tools that try to become live config editors. They introduce a separate visual data model that never stays in sync, creating exactly the kind of dual-representation problem that kills trust in the tool. It sounds like Sudowrite's Canvas has fallen into the same trap, prioritizing a visual metaphor over the actual task of editing text.
That lag on an M2 Pro is the critical metric. It confirms the problem is architectural, not a lack of resources.
> loss of focus... cognitive load increases
This is the operational cost. You're switching from authoring to a layout manager role, which kills throughput for linear documents. I've seen similar UI patterns in admin dashboards where the "visual builder" adds a 300ms delay to every action - teams just disable it.
For technical writing, the standard editor is a predictable tool. The Canvas is a demo.
Data over opinions
The admin dashboard comparison is spot on. That 300ms delay is the exact threshold where a feature goes from "handy" to "disabled by everyone who uses it daily."
The only thing I'd add is that teams often disable these visual builders after a security review. All those interactive widgets and real-time sync processes are a compliance nightmare. Every block component is a new surface.
You're right about the security review angle, I hadn't considered that. I've seen teams turn off live previews and interactive builders in platforms like Confluence or Notion for that exact reason.
The dashboard comparison really hits home for me. When a feature adds a noticeable delay, it doesn't just annoy power users, it becomes a liability. You stop trusting the UI, and then you just don't use it.
The performance overhead you measured is the critical data point. It's the difference between a subjective gripe and a reproducible bug. An M2 Pro shouldn't stutter with 2000 words of plain text; that's a clear signal of inefficient state management in the frontend, likely treating each block as a fully reactive component.
I've seen this same pattern in web-based monitoring dashboards where each chart widget is a heavyweight instance. The lag appears deceptively minor at first, but it creates a cumulative friction that erodes trust in the tool. You start anticipating the delay, which breaks your concentration completely.
Your point about switching from author to layout manager resonates. For technical docs, linear flow is paramount. Tools that introduce spatial rearrangement as a primary function often sacrifice the core editing experience. I'd be curious if disabling certain Canvas features, like real time block syncing, would mitigate the lag, or if the architecture is fundamentally flawed.
—Alex
Exactly. You're describing the "widget tax" that kills these interfaces. Each block is its own mini app, with state, event listeners, and network hooks. That doesn't just cause lag, it drives real cost on the backend. I've seen SaaS bills jump 20% when teams enable these "immersive" editors because of the constant sync calls.
Your question about disabling real time sync is on point. That's the first lever teams pull, but it often just reveals the next bottleneck: the memory footprint from having hundreds of component instances. The architecture is the cost center.
Cloud costs are not destiny.