Comment by stouset | Hacker News Reader

Comment by stouset | Hacker News Reader

Parent Full thread

> No matter where I look, up and down the stack, across different OSes and tech stacks, there are bugs.

I’m not sure I’d go quite as far as GP, but they did caveat that we often choose not to write software with few bugs. And empirically, that’s pretty true.

The software I’ve written for myself or where I’ve taken the time to do things better or rewrite parts I wasn’t happy with have had remarkably few bugs. I have critical software still running—unmodified—at former employers which hasn’t been touched in nearly a decade. Perhaps not totally bug-free, but close enough that they haven’t been noticed or mattered enough to bother pushing a fix and cutting a release.

Personally I think it’s clear we have the tools and capabilities to write software with one or two orders of magnitude fewer bugs than we choose to. If anything, my hope for AI-coded software development is that it drops the marginal cost difference between writing crap and writing good software, rebalancing the economic calculus in favor of quality for once.

> I’m not sure I’d go quite as far as GP, but they did caveat that we often choose not to write software with few bugs. And empirically, that’s pretty true.

Blame PMs for this. Delivering by some arbitrary date on a calendar means that something is getting shipped regardless of quality. Make it functional for 80% of use, then we'll fix the remaining bits in releases. However, that doesn't happen as the team is assigned new task because new tasks/features is what brings in new users, not fixing existing problems.

I don’t disagree but is the alternative unbounded dev where you write code until it’s perfect? That doesn’t sound like a better business outcome. The trade off can’t be “take as long as you want”

Some coding doesn't fit your schedule. If you've scheduled 2 weeks, but it takes 3, then it takes 3. Scheduling it to take 2 does nothing to actually make the coding faster.

3 sounds fine.

Then I ask: why not add a week to how long that thing will take, meaning it stretches two sprints (or whatever you call it).

Add upfront. Then if you get to hard convo where someone says “do it sooner” you say “not possible.”

The fundamental problem remains: it’s difficult to predict how long it will take to solve a series of puzzles. I worked in a dev group where we’d take the happy path estimate and double it… it didn’t help much. So often I’d think something would take me a week, so two walls was allotted, but I made a discovery in my first like hour/day whatever that reduced the dev time to like a couple days. Then, there were tasks that I thought I’d solve in a few days that took me weeks because I couldn’t foresee some series of problems to overcome. Taking a guess and adding time to it just shifts the endpoint of the guess. That didn’t help us much.

That's the point I am making, and the point of asking "what is the alternative"

Developers aren't alone in adhering to schedules. Many folks in many roles do it. All deal with missed deadlines, success, expectation management, etc. No one operates in magical no-timeline land unless they do not at all answer to anyone or any user. Not the predominant model, right?

So rather than just say "you can blame the PMs" I'd love to hear a realistic-to-business flow idea.

I am not saying I have the answers or a "take". I've both asked for and been asked for estimates and many times told people "I can't estimate that because I don't know what will happen along the way."

So, it's not just PMs. It's the whole system. Is there a real solution or are we pretending there might be? Honest inquiry.

Software release dates are so arbitrary though. We no longer make physical media that needs time to make and ship. Why does software need to be released on February 15th instead of March 7th?

You could ask the same question about the contents of the release. Why does software need to be released with features X, Y, and Z on March 7th when it could be released with features X and Y on February 15th?

It's inevitable that work will slip. That doesn't necessarily mean the release will slip. Sometimes you actually need the thing, but often the work is something you want to include in the release but don't absolutely have to. Then you can decide which tradeoff you prefer, delaying the release or reducing its scope.

This is the direction of my thinking, too.

Earlier discussion focuses on writing software at a slower pace to inject more accuracy and robust thinking/design/code. Conceptually, yes, I get it!

But in numerous practical scenarios, some adherence to a recurring schedule seems like the only way to align software to business outcomes. My thinking is tied more to enterprise products (both external and internal) rather than open-source.

I like an active dialog with engineers. (I'm neither SWE nor PM). Let's talk together about estimates. What's possible and not possible. Where do you feel most uncertain, most certain. What dependencies/externalities do you expect to cause problems.

Those conversations help me (business/analytics-side) do things like adjust my own deadlines, schedules. Communicate with c-suite to realign on what's possible and not. Adjust time.

The main problem I’ve had is the unpredictability of where the complexity lies. Unless you’ve done exactly what you’re doing, before, with the same tools and requirements, there’s a good chance that some discrete trivial aspect could take up an incredible amount of time, and that won’t indicate whether the main goal will take more or less time. I’ve worked both as a developer and as a designer, and while some aspects of design can be really nebulous and uncertain compared to dev work, it lacks some of the unpredictability — it’s not like I’m going to unexpectedly have to re-make the logo.

I feel for anyone that has to wrangle these tasks into a business-consumable time frame.

> The main problem I’ve had is the unpredictability of where the complexity lies. Unless you’ve done exactly what you’re doing, before, with the same tools and requirements, there’s a good chance that some discrete trivial aspect could take up an incredible amount of time, and that won’t indicate whether the main goal will take more or less time

What a great articulation. Completely agree.

This is why I don't blame PMs anymore than devs anymore than business folks throwing requirements at PMs. Possible to find fault everywhere.

I think the broader problem is scale and growth. Many people in many roles are caught in growth-mind or scale-mind companies where the business wants to operate at a velocity that may not align with the realistic development work we're discussing. PMs are similarly caught with less time to understand, scope, plan, etc. Business folks ask questions like "why isn't this ready" to devs that may not understand the reasons why the business operates the way it does, or the business at all.

Full disclosure: I'm in insurance. Seeing lots of these problems play out in front of me. C-suite moving at speed 100, devs moving at a perceived speed of 50. Silos and communication problems and unclear requirements up and down the stack.

So, in my interactions, the way I try to help is just to understand the most basic components and their ability to come alive or not. Is there anything to show? Yes, ok - let's celebrate a small win. Is there a rather large delay? Why - ok, let's use that to reinforce building something robust vs. crap.

But, there are schedules! Someone above mentioned sqlite. Another example comes to mind: Obsidian. I think they're anomalies (good ones) rather than examples that broadly prove the point to slow down.

> Why does software need to be released on February 15th instead of March 7th?

Because it has to be released at some point, and without picking a point in advance, you can never reach it.

https://en.wikipedia.org/wiki/Parkinson%27s_law

I disagree with that entirely. Some features just take longer to develop. If that feature is part of the release, then release it when it is finished and not kind of working. If that feature is just not achievable, then PMs have really screwed up their role by putting it in the release in the first place.

You assume that PMs will just accept whatever estimate you give and not just say 2 weeks from the off and refuse to budge.

So, could you say "ok, but I still can't do that"

In this day and age of code-in-bulk enabled by AI, they will find someone who does in a blink of an eye.

“The alternative is that nothing will ever get released because devs will take forever making it perfect” is a really lame take.

We have literally countless examples of software that devs have released entirely of their own volition when they felt it was ready.

If anything, in my experience, software that’s written a little slower and to a higher standard of quality is faster-releasing in the long (and medium) run. You’d be shocked at how productive your developers are when they aren’t task-switching every thirty minutes to put out fires, or when feature work isn’t constantly burdened by having to upend unrelated parts of the code due to hopelessly interwoven design.

I'm happy to be reoriented with examples. Please provide some? You said countless but mentioned none.

tex was pretty bug free.

I'd say SQLite is one good example:

https://sqlite.org/chronology.html

Regular releases for over a quarter of a century now, and it's renowned for its reliability.

I think PMs fail to understand categories of change in terms of complexity because they focus on the user facing surface and deal in timelines. A change that brings in a big feature can be straightforward because it perfectly fits the existing landscape. A seemingly trivial change can have lot of complexities that are hard to predict in terms of timelines.

There is also the angle of asking for estimate without allocating time for estimation itself.

For lack of a better word, I think it should drive from "complexity". Hardness of estimate should be inversely proportional to the complexity. Adding field to a UI when it is also exposed via the API is generally low complexity so my estimate would likely hold. We can provide estimate for a major change but the estimate would be soft and subject to stretch and it is the role of the PM to communicate it accordingly to the stakeholders.

Hearty agree. I think the PMs fall victim to wildly optimistic imagination of how fast and easy it will be to correct from “good enough to not get yelled at by CEO for not shipping by X date” to “works correctly and isn’t creating more bugs” - and importantly, it seems like they repeat this mistake every project, compounding the problem. So we perpetually have an increasing number of hacks, interacting with each other to cause difficult issues, all of which the PM says we will fix next sprint, just as soon as we ship one more Important Feature.

Not all orgs of course. But most I’ve personally seen, seem to be like this.

Reply on news.ycombinator.com