Hacker News Reader

Comment by pharrington | Hacker News Reader

Professional AAA game programmers write bugs that permanently delete players' save data. Senior engineers at web hosting companies accidentally delete customer's entire servers. Employees at internet backbones accidentally screw up the configuration so bad that huge swaths of the internet go offline.

This isn't a UX problem. This isn't a government problem. This is the human experience.

View on HN

More like, this is the state of software in 2023. In software industry, there is an extremely vague sense of correctness while software is being developed, if there is any. Product managers have no idea what that is, and there is usually no specification to test or verify software against. Every engineer has their own voodoo ways to test or verify software, and guess what, some teams don't even do either! There are entire companies, philosophies built around how testing and verification being waste of time. On top of this, software depends on gazillions of dependencies whose correctness is similarly untested and unverified.

As a corollary, you get software that fails left and right for unexpected reasons.

Yeah people don't give a shit about correctness. By people I mean PMs. And it seems fine for a long time until things get ruined like any of those examples. "How could this have happened?" said the people who refused to accept that testing can added tens of engineer-days to the estimate.

You can only ask a PM "how would you like the system to behave in this specific scenario" so many times and get non-answers or no answer before you start saying "screw it, I think it should behave like this, ship it"

My favorite response is "that feature's deprecated"

Yeah....I know...but it still exists in prod and it's going to be affected. "Deprecated" does not mean "obsolete".

And yet, extraordinarily complex systems seem to work somehow, which means each of the zillions of pieces is pretty much doing the right thing.

That's your opinion, and it seems to be the opinion of some people, I respect that. However, I strongly disagree. I use software all day every day (like most of us here, if not all) and I can confidently say none of the complex systems I use work at all. All software is buggy, most software is frustratingly buggy, and surprising amount of professional software is imho unusuably buggy. When I interact with a new software, my assumption is that it will not work. Unless proven otherwise, software of no complexity works in practice.

Unless you are an exceptional person, software has permeated your life. Most of the ways in which it affects you are invisible because it works so well.

My claim is not that "no software has any part that works", my claim is that "all software has some part that is buggy" which seems trivially demonstrably true in my life. I have unfortunately nothing to offer but anecdotes, but I go by my day and almost every moment something fails. I open jira tons of errors. I open an Android app, tons of issues. I open LinkedIn in browser, sometimes text box doesn't work. I open facebook, it has gazillions of weird problem that pop up every moment.

In fact, I'll go against the wind a little more. A sibling comment in this chain claimed when software has bugs, it's resolved quickly. Wow, I cannot disagree more. I know bugs in tons of software I use that's been there for years, since I started using it. In fact, my principle is the exact opposite, if a software has a bug, unless you yourself fix it [1] chances are that bug won't be fixed.

[1] if open source

>... software of no complexity works in practice.

That's trivially disprovable by mere fact that you have used software to send your opinion to everyone this forum also using untold amount's of software, and that we have received it and are replying to it. Unless you are using some black and white definition of `works` where everything acts exactly as expected 100% of the time, in which case I would suggest not doing anything ever because there is no agreed upon definition of behavior on anything more complex than basic math. Once you start defining behaviors there are a million ways to interpret any sentence

"none of the complex systems I use work at all"

None of them?

I think by far the best working software I use today is Emacs, and yes it's full of bugs that frustrate me every day.

That's the crazy thing, right?

I'm a software engineer. I've seen how poorly software is written, how badly it's tested, and how many stupid mistakes developers make. I've been the one making staggeringly stupid mistakes!

And yet... the vast majority of the software we use just kinda works. And if it's broken, it doesn't tend to stay that way for long.

It's easy to see problems as a developer, but our software could be so much worse.

I think I'd push back a little bit on this. Watching non-technical people try to use apps and websites makes me feel an immense amount of guilt. It's crazy the amount of problems that I ignore because I am so used to them or because I understand the inner workings of things. When a non-technical person sees one of those problems, it's sometimes a show stopper. It wastes a lot of brain cycles for them to figure out what error messages mean, why there's a pop-up explaining some new arcane feature unrelated to their goal, why their form submit didn't go through, etc.

The fact that software is so pervasive definitely speaks for itself, but we still have a very long way to go.

I can't disagree with you about that. The article here is about a misconfigured zoom meeting, which is shockingly easy to do. And if you've ever seen technical people struggling with Jira, you'll know it's not just an issue for non-technical users.

But I think it is a different quality of the software. Contrary to other people in these comments, I think we make software that actually does work in the vast majority of cases. "Working" isn't really the same as high performance, feature rich, or intuitive software, though.

> why there's a pop-up explaining some new arcane feature unrelated to their goal

This is one of the worst things apps and websites have started to do. I can hardly open an app without encountering this. Note to devs and PMs who think these are a good idea: I don’t read them, I dismiss them immediately, and every time you do this I like your app a little bit less. Stop it.

My favorite thing is gmail or chat telling me about some new feature using an overlay instead of, I dunno, emailing it to me or sending me a chat message.

It's gotten ridiculous. I log into my bank and click on my checking account and instead of taking me to a page with all my recent transactions it instead takes me to some page to set up savings buckets or something that I'll never use.

My internet was down the other day so for something to do (I didn't think it would actually work, and it didn't) I logged into Comcast's site on my phone to see if there was an outage or something I could troubleshoot from there. I had to click through maybe half a dozen notifications of things I don't care about in the slightest just to get to the "home" page of my account.

And that's just the first two things I thought of. That sort of miserable design where they try to get me to look at things I don't care about and will never care about just because it might make them money is so pervasive, it's everywhere.

A lot of the software you interact with on a daily basis is made to be robust at least from a system point of view. Firmwares are tested because if they fail horribly the product will get returned or cause real damage. Communication protocols are designed to be fault tolerant. Critical infrastructures operate far from their peak capacity and are doubled. There are buffers everywhere limiting the impact of a brittle piece.

People seldom realise how much engineering is necessary to make the modern world tick and that goes for software too.

Until the moment you do anything weird or unexpected, then it goes haywire

As someone who hates "average" and "normal", this is a big problem, particularly when the goalposts on that have moved to pretty much anything that anyone wants to do beyond elementary-school education-level difficulty

The ease-of-use assholes that have ruined software can kindly go eat a dick

This is life.

People fail to recognize so much of life is imperfect down to our own biology and the design of the systems that span our planet as well.

At some point, there is no return on the investment of perfect design.

It seems our entire world is underpinned by “perfect is the enemy of good.”

- https://cosmosmagazine.com/science/human-egg-cells-are-imper...

- https://biology.mit.edu/revealing-an-imperfect-actor-in-plan...

- https://news.harvard.edu/gazette/story/2020/04/human-knee-ev...

Most complex systems don't work correctly. They just don't fail catastrophically very often. They fail in small ways instead. A lost user account, an incorrect payment, a missing parvel, a poorly routed call. These things happen regularly - millions of times a day in fact - but out of hundreds of billions of interactions its hard to notice them.

We're so used to things failing we stop noticing and think slightly broken is the same as working properly. That's kind of OK until it's your payment, or your parcel, or your account.

> More like, this is the state of software in 2023.

I don't believe it's ever been any different.

Maybe we can improve the human experience by improving the… user experience.

Exactly. It doesn't matter if UX was the problem, it can be part of the solution.

You're very right. I may have misspoken a little - I just meant that there will always be high profile mistakes. Bu obviously that's no reason to avoid improving the tools.

In what roadmap did I block the deliverable of my own happiness?

https://youtu.be/DYvhC_RdIwQ

you can lock meetings that has started

this is not a UX issue, it is people who give no fuck about security of privacy

otherwise your government would have voted anti data-spying laws, they haven't therefore they don't care, therefore they didn't think of locking their meeting after it started

i can spell the agenda that's cooking

https://blog.zoom.us/keep-uninvited-guests-out-of-your-zoom-...

https://support.zoom.us/hc/en-us/articles/201362603-Host-and...

It's not that hard to not fuck up that bad. It's hard for PMs to accept the cost of it preemptively though.

Somehow thousands of passenger flights happen every day and the planes don't fall out of the sky.

Perhaps it's possible to do better.

How old do you suppose that software is?

There are hundreds of different pieces of software used every day to keep planes flying, both on the aircraft and on the ground.

Given what it costs to make software for planes, maybe the status quo isn't so bad.

And yet someday we may look back and see this as the good old days.

And soon to be the AI experience, trained upon fallible human experiences.

Human factors

Reply on news.ycombinator.com