Thank you so much for running this needed call. It's nice to see that the concerns around the DHF and transparency issues are being taken seriously, especially with @blocktrades getting in on the convo.
I posted something in Discord, but I am going to post it here too concerning testing. I've got a lot of testing experience across multiple types of testing:
With computer/tooling use, AI can do this increasingly well now. I have an Android/iOS game I built that had a pathfinding bug. I got Codex with GPT-5.4 to open up the iOS emulator, click around, debug console output and fix the issue (so you can do semi-automated mobile testing).
Combined with established testing methodologies:
- Unit tests
- Integration tests
- End-to-end tests
- Smoke tests
- Visual regression testing
The tooling side of things was already resolved before AI and now AI is making testing even easier. It's no longer a chore or PITA to set up good test harnesses.
I'm honestly a little confused why mobile testing is allegedly a sticking point for time/cost for projects, because mobile emulation isn't hard. The iOS emulator supports numerous emulations, same with the Android emulator. Maybe this comes down to inexperience and people who aren't mobile devs are building mobile apps?
Also, building mobile apps is easy now too. React Native is a mature ecosystem that lets you build cross-platform mobile apps from the one codebase. There aren't really any reasons for people to develop native mobile apps anymore. React Native + Expo is incredible.
It's incredibly easy to build apps on Hive now. I've even got a simple starter I open sourced on GitHub called Honeycomb: https://github.com/Vheissu/honeycomb which gives you an instant starting point for Hive applications (Node.js backend, indexer, DB integration and front-end). I also have a library I've been working on for the past 7 years for Hive called Hive Stream: https://github.com/Vheissu/hive-stream which gives you a simple, extensible and adapter driven approach to working with the blockchain (supports MongoDB, Postgres, SQLite, MySQL).
From my perspective 80-85% for developer expenses is far too high for a project to spend its money on. It signals that for a user-facing project, that it has no real go to market strategy, no one is doing user research or design properly, QA is treated as an afterthought (or done by the devs which means dev work stops) or worse: the team is just larger than it needs to be.
What is the output relative to the spend? A solo dev shipping a game every couple of weeks makes the math look pretty brutal for a funded team burning 85% on developers and shipping quarterly at best. The ratio itself isn't inherently wrong, but it becomes a red flag when velocity doesn't match the investment.
AI tooling has genuinely changed what a small team or solo dev can produce (as everyone on the call acknowledged, with figures cited around 10x more efficient). If someone is employing the same team size they had three years ago and pointing to the same developer cost ratio, the honest follow-up is: what are all those people actually doing day to day? Because the productivity ceiling per developer has gone up significantly.
Maybe it's a visibility issue, but I don't see many Hive projects that are DHF funded shipping any faster than they were prior to AI tooling becoming good.
One of the big questions I want answered is, what the heck are Hive Keychain spending 600 HBD a day on? Because another takeaway from this call (that I have already mentioned numerous times in my posts, comments and Discord messages) is infrastructure costs for Hive are dirt cheap. Hence why projects on this call all seemed to cite close to 10% ish of funding for infrastructure, which anyone who works in tech and has experience with servers already knows (especially when your apps are offloading the database work to the chain itself).
Thanks for the thoughtful comment - a lot of what you said about testing tooling and AI improvements is absolutely true.
Part of comment related to Ecency mobile app, so I will leave some more notes about our process. Where things differ a bit for production apps like Ecency is that emulators and automated testing don’t fully represent real-world mobile environments. Android alone has thousands of device/OEM combinations, and issues often appear only on specific real devices (memory limits, background restrictions, camera APIs, push notifications, etc.). These are difficult to reproduce purely through emulation. You cannot avoid using emulators/simulators on mobile development, it is done already since inception of the app.
Another thing to add, Hive apps add extra complexity: wallet signing, broadcasting transactions, interacting with multiple RPC nodes, and integrating with tools like Keychain or HiveAuth. Many of these flows involve real blockchain interactions that can't always be safely simulated.
Our code is based in React native, in practice there are still platform-specific bugs, native modules, OS changes, and app store constraints that require additional testing and coordination before releases.
Automation, AI, and good test coverage definitely help - but with a production app used by many people (and involving crypto transactions), stability and careful releases still require real-device testing and QA cycles. So we are exploring how to improve those and optimizing.
Appreciate the discussion and ideas around improving development efficiency. Looking forward to test your Android/iOS game, please do share link/listing.
GitHub is not working 404 error
haha, oops. Forgot to make the Honeycomb repo visible. It's all good now
Your reply is upvoted by @topcomment; a manual curation service that rewards meaningful and engaging comments.
More Info - Support us! - Reports - Discord Channel