# David Torrance, Independent Developer (ex-editor) — read of iroh-p2p-video-collaboration, June 21 2026

> Nine years cutting video at a mid-size commercial house, left in 2022 to do contract React work. Still have opinions about frame.io. Currently trying to find something worth building that I actually understand.

## How I got here

Searched "video collaboration tool indie hacker" after a post in the Indie Hackers forum mentioned someone trying to build in the post-production tools space. Page came up third or fourth in results. I clicked because I've thought about this specific problem for years and wanted to see what someone else's angle looked like.

## What I clicked first

The hero pulled me in for about eight seconds: "Edit together, anywhere. Peer-to-peer remote video collaboration. Real-time sync. No servers. No latency."

I stopped on "No latency." That's not a real claim. Video files are large. P2P connections go through whatever ISP your collaborator is on. I've used Evercast, I've used frame.io's review suite, I've dealt with editors in Manila connecting to a project in Austin. "No latency" is a thing you write when you want to sound impressive and you haven't talked to anyone who does this for a living.

## Where I paused

The self-scoring table. "financial upside: 1/10" and "market openness: 4/10" and then a projected Year-1 take-home of negative $33,948. I stopped and read it twice because I thought I misread it. The page is trying to sell me this idea and it's also telling me, in its own words, that the financial upside is one out of ten.

I actually respect that they printed it. Most idea marketplaces bury the bad news. But I'm now trying to reconcile the top of the page ("The future of video collaboration is distributed. Iroh proves it works today.") with their own admission that the financial upside is nearly nonexistent.

## What I distrusted

"Iroh proves it works today" is doing a lot of work for a concept that has, by the site's own disclosure, zero live customers. That phrase belongs on a product with a demo, a GitHub, a video of the sync actually happening. There's none of that here.

"All edits are cryptographically signed to prevent tampering" is the kind of sentence that sounds technical enough to skip over, but I'm a developer and I want to know what that actually means at the implementation layer. Is this CRDT-based? Is timeline data structured or just a diff blob? What happens to a six-terabyte RED project on asymmetric fiber? The "How it works" section describes "distributed networking protocols" which is not a sentence, it's a placeholder.

The scoring axes gave "buyer clarity: 10/10" and "credibility: 9/10" and I'm not sure I agree with either. The buyer isn't that clear to me. Is this for solo filmmakers or broadcast facilities? The page says both, and those are wildly different problems.

## What would convince me

A five-minute screen recording of two editors on different ISPs actually syncing a timeline. Not a marketing video, a raw screen capture. Show me the latency number in real conditions. Show me what happens when one person makes a cut while the other is offline.

Or, if that doesn't exist yet, then at least a link to a technical writeup explaining how NAT traversal is handled and what the sync protocol actually is. The name "Iroh" is the name of a real open-source Rust library for P2P networking by n0computer. If this is built on that, say so. That would be meaningful signal.

## What I'd ask in an email reply

1. Is there a working prototype? Even internal, even rough -- can I see timeline sync actually happening between two machines on different networks?

2. The market openness score is 4/10 and you didn't elaborate much. Frame.io is owned by Adobe. Evercast exists. Sohonet exists. LucidLink is growing. What's the specific wedge here, and who is the one buyer type most likely to pay for this in year one?

3. The $99 tier includes "working code starter" -- what does that mean? Is it a repo with the networking layer wired up, or is it boilerplate React with a readme?

## Verdict: on-the-fence

The problem is real, I've lived it, and I appreciate that they printed the bad Fermi numbers instead of hiding them. But "No latency" in the hero and "1/10 financial upside" in the scoring section is a dissonance I can't quite get past without seeing something that actually runs.

---
*Memo by skeptic persona, generated 2026-06-21. Studio breaks own self-grading loop.*