Remix.run Logo
▲ bengold14 5 hours ago

I think most people at this point just don't review code. Which is a shame, because:

1. It was good for collaboration. Sharing how you are changing the codebase pushes everyone to row in the same direction, and provides a sense of team ownership.

2. It was good as a push for engineers to take pride in their work. Knowing someone is watching gives you subconsciously an extra reason to not cut corners. And when you spend effort not cutting corners you expect that in your teammates, which is a (potentially) virtuous circle.

3. It was good for catching mistakes. When code was expensive this was important.

Everyone thinks of code review as primarily relevant for 3, and since code is no longer expensive they think it is unimportant and you can skip it. But 3 isn't dead, its importance has just shifted from code to architecture. It is still an expensive mistake to screw up architecture and have other engineers build upon that flaw. Re-architecting with live users, data and/or hardware is not as simple as fixing a code bug, and it deserves attention to catch mistakes. Collaboration & pride are also still important.

This is the reason that instead of stopping review, we need to change what abstraction we review. Code is no longer the right tool for understanding applications in most cases.

(disclosure I work in this space and run https://www.archme.io which is working to solve the above)

▲cgarvis 5 hours ago | parent | next [-]

I agree with your premise but I would argue that your still operating under the model that we are all still working with all the "developers" that we needed before because we need someone to type out the code.

I like your ArchitechMe approach because it operates at what is left. The Software Engineers still need to "work on the system, not in the system". Before those folks were in the code, but with so much being written now we need better tools like you ArchitechMe to make sure the system is operating like we want it to.

▲bengold14 5 hours ago | parent [-]

Thanks for taking a look! FWIW I believe that we'll still need engineers and more of them, because we're going to start to build more complicated software, faster.

The problem right now (and why teams are shrinking) is because people haven't figured out how to collaborate at scale with AI IMO, so they just give up and say it's easier to manage a small team of seniors.

▲matltc 4 hours ago | parent | prev | next [-]

What's up with all the trackers? Sounds interesting but closed immediately due to their presence

▲bengold14 3 hours ago | parent [-]

Apologies for the trackers - it's just vercel & posthog so we can improve the website and app.

Is this not common practice?

▲esafak 4 hours ago | parent | prev | next [-]

I don't agree with your premise, that code is cheap and unimportant. It is cheap, but bugs aren't unless you have a monopoly, and vibe coded software as you describe is invariably buggy.

▲bengold14 3 hours ago | parent [-]

I'm not saying it is unimportant (it's the basis for everything), but I am saying that agents do a bang up job of writing it at this point, and to get the benefit of agents you need to stop reading the code.

It's almost has hard to have agents write the code and review every line than it is to write it yourself IMO. So either they write and your review and you get a 10-20% bump, or vibe engineer and get a 200% bump.

I'm attempting to build a way to avoid looking at code, while still being able to engineer because frankly it's hard to engineer without looking at a shared artifact. Anthropic et al want us to just stop engineering and let agents take over, I think we should be empowering humans to work at the correct level with the agents instead.

▲esafak 15 minutes ago | parent [-]

Claude Code has over 5,000 filed issues, and other vibe coded repos are similar.

What are we bumping 200% exactly, the PR rate? What about the bug rate, and volume?

▲imtringued 3 hours ago | parent | prev [-]

I'll be honest about your project. Asking an agent to create the model is the wrong way around.

It may surprise you but code review is actually something you need to automate with deterministic software.

For some reason people don't seem to understand what a developer's job is.

Automating code review sounded asinine two years ago because the cost of managing the automated process exceeded the cost of manual reviews. Aka it was pointless busy work.

But now that we have potentially adversarial agents, it is kind of essential.