Long ago I created a Chrome extension that reordered things nicely, but their CSS changes frequently which made it a maintenance hassle, and I read on HN another dev saying doing so is against their terms of service anyway. Not a fan of that anti-tinkering attitude.
SLACK: don't make dramatic changes to a user's workflow without giving a simple toggle to preserve old behavior.
I get it, most users probably love this feature. My wife does, for example, and she works at a big organization, which has different needs from my workplace. But even if dramatic changes to product are approved of by 75% of users, every time you do so, you create whiplash for the other 25%, and prevent many from ever loving your product. Rinse and repeat that whiplash too many times and product design rants on HN with 600+ upvotes will be a regular occurrence...
Translation: "we're paying all these engineers and product managers and they need something to do"
Instead they spent actual time, effort, and money making their product worse.
My company has certain deficiencies, but one of our core principals is that we'll never break a workflow, even if the workflow is dumb, even if the "feature" is actually a bug that an enterprising user abused in a way we didn't anticipate. The bad news is that we're saddled with a ton of legacy crap that can't be rewritten. The good news is that we've grown into one of those behemoths that dominates a niche specialized industry and won't be unseated by a product that is only, say, twice as good as ours. It's not as fun as iterating fast and breaking things, but the low stress is nice.
Ideal software would be continually fine-tuned and shrunk -- there'd be no bi-annual massive redesign, no change for change's sake alone. Instead of bored devs sitting around an office looking for ways to integrate $FRAMEWORK_OF_THE_MONTH and get those coveted resume points, a well-run project would make something work well and then they'd leave well enough alone, focusing only on bugfixes, performance, and other "boring" projects that don't make for big press releases. Changes to working products should be as surgical and minimal as possible.
A good compensation structure that would prioritize stability and consistency would pay an ongoing royalty to the relevant technical people based on the product's performance, uptime, and minimal crash/bug occurrence. "Hours worked" would be minimally relevant. One wonders if so many people would be so desperate to desecrate their production infrastructure if a high-quality work product and compensation were actually correlated.
But because we can't break out of the assembly-line 40-hours-per-week mentality, we pay developers as if they're line workers, and there's always got to be something on the line to keep those worker bees buzzing, regardless of the aggregate negative impact of constant uncoordinated meddling in complex systems.
This hits painfully home for me. I learned a few weeks ago that some of my coworkers did something akin to this. They were working on what was frankly a mostly-silly project to preserve the relevance of increasingly irrelevant internal tools. Once they had produced something working, they stopped. Then they re-implemented the whole thing in Rust.
Which few in the company know or use. There are no clear benefits to this except exciting resume points for the developer in question.
That was discussed on HN in 2014[1] and right in the top comments is "I think one factor that leads to bloated, ruined software was missed... I don't know how common it is overall, but I have personally seen it ruin several very good products. And that is the simple fact that employers want their employees to remain busy. If a piece of software reaches a point of exceptional quality - the developers working on it still have to fill 40 (likely more) hours a week to appease bosses. And so they do the only thing available - they ruin the product."
No, they didn't. They just stopped typing. Hence, bullshit UI's like Instagram.
> Steve Jobs knew it beat a physical keyboard
Yes, if by 'beat' you mean 'made it too expensive for general use'. But in general it was a huge blow to usability, productivity and general global intelligence.
Your mistake here is confusing a consumer device with something designed for a professional workflow. Touch screen keyboards are, objectively, not faster to type on than a physical QWERTY keyboard. It's like arguing that because my mom got used to a using a cellphone for photography that physical lenses are obsolete.
Again, this is (in my mind) something that mostly only happens in the software industry. A professional racer drives with a manual transmission. When automatic transmissions came out, nobody argued that professional racers were holding back the auto-industry. Meanwhile in the software industry, mention that your workflow benefits from corded headphones, and suddenly you're jeopardizing the future of progress.
I cannot wait until the MBP 2025 with the whole keyboard replaced by the touchbar..
Who needs tactile feedback?
First, is it possible to develop software without inconveniencing business users with temporary (let's assume the ultimate products are better then the linux/vim/emacs they are superseding) regressions and inconveniences? And what is the cost in terms of time to route around such pitfalls? Would we still be able to have startups at all if products required thousands of hours of QA or perfect test suites in order to launch?
Second, if we were to set a hard rule 20 years ago that all software was to avoid this phenomenon during its development, what valuable tools and services would never have been developed at all? Would we still have Twitter? Reddit? Steam? Whatsapp? I don't have to dig far into the history of any of those tools to find near revolutions by their userbases over braindead UI or adversarial practices in the name of "vision".
I don't know, these are open questions. I just think avoiding all such frustrations you mentioned is wishful thinking and at some point it is just part of the process of experimentation and iteration. Or perhaps this process is entirely different in a small corp. versus a big corp. environment.
How many of those broke old workflows to introduce new ones with no clear benefits?
But that's not the point. We're not talking here about broad services like Reddit, Twitter, or Steam. We're talking about something more akin to a libary. You shouldn't break interfaces in a library without a strong, compelling reason.
Do things need to change eventually? Absolutely. Do they need to change today, because someone decided that the old workflows they don't like need to break for everyone? Maybe not. There's perhaps some room between the two.
It's been my experience that business software often represents a deep investment in a given workflow. Sometimes to the point where businesses are willing to spend a great deal of money to preserve those workflows and integrate new things into them - MuleSoft springs to mind.
Which is not to say that you're wrong. It's absolutely possible to develop new, improved software that's better in critical ways. Sometimes people and businesses are willing to put up with temporary regressions and inconveniences to gain substantial improvements. To use the above comparison, I've seen artists invest the time in learning how to draw all over again in order to jump from paper to digital.
What Slack has done is take away what was a perfectly functional workflow for many people. This doesn't look like a temporary regression or inconvenience. This looks like a permanent, hard break without substantial obvious benefits for people who used the old workflow. Communicating with users in a way that telegraphs very clearly that Slack doesn't care at all is just gilding the lily.
I can stay in older functional versions for a long time without being forced to upgrade and disrupt everyone's workflow because someone in a company thought they knew better how we should work.
Jenkins, Gitlab, rocketchat, review board all open source tools that you can be running years old ( on an isolated network please ) without upgrading and being very functional...
The trick is to avoid introducing change without simultaneously introducing perceived value to the user. Predictability and reliability are incredibly important for business users, because they're going to build processes and documentation and training around however they interface with your system. This includes both directly interfacing with your system and creating supplementary processes outside of your system to work around deficiencies your system has in supporting whatever need they have. If you introduce change, you inherently reduce the temporary predictability and reliability of the system, and the friction created for end users needs to be perceived as worth whatever value your change provides. If they perceive a benefit to themselves, it'll incentivize them to embrace and adapt to the change. If they perceive it as a regression to their processes/needs, they'll fight it and it'll eventually lead to resentment and friction between the business users and the developers.
Slack's WYSIWYG input box is a prime example - it unlocks capabilities which is convenient from Slack's perspective, and is potentially generating value for users not familiar with Markdown and are used to traditional GUI-based rich-text editors. Helping them better support the type of corporate-wide adoptions that are lucrative for them. But due to Slack's origins, a large chunk of their user base prefers the older, Markdown-based editor and perceive a decrease in value from the change. If you're going to introduce changes such as this which will be perceived as a regression by a large enough subset of users, then you need to account for that in your change management process such that it isn't abrasively disruptive.
What happened to Digg? Myspace? UI isn't the only reason, but it's definitely a contributing factor in their demise.
The eBay story about changing the color of the banner should be followed for new features as well.
Imagine that every change you're making is like hitting your user in the face with a brick. If you're going to give me something amazing that makes my life better, I may let you hit me in the face with a brick so I can get it. But if you hit me in the face with a brick and then you give me something worse than what I had? Don't do that.
Learning the basics of Vim was a really big, hard change for me, but it was an adjustment to my workflow that was made for a specific reason, that made my life better and that made me more productive, and that (importantly) was a decision I made voluntarily. It was not change for its own sake.
> Would we still be able to have startups at all if products required thousands of hours of QA or perfect test suites in order to launch?
On the contrary, how many more interesting, better chat apps would we have if every one didn't feel the need to reinvent Markdown? Wouldn't it have been more useful if instead of rebuilding their editor for no reason at all, Slack's engineers instead added new API endpoints, or experimented with encryption, or added new search tools, or addressed any of the pain points that actually get raised by professionals using their product?
I would also push back against the idea that this is simply the cost of experimentation. Vim/Emacs are much more experimental editors than Word, yet even modern remixes of those editors like Spacemacs put more thought into user customization and consistency than Word does. Spacemacs' keybindings evolve -- but they never force you to accept that evolution if it would break something fundamental to your workflow. And Spacemacs is doing way more experimental, interesting stuff than Slack is.
To add further onto that idea, prioritizing future innovation over the productivity of real users is a very software-specific philosophy about how a professional field should work. New animation techniques come out all the time, but we don't look at people like Miyazaki and say, "that man is holding us back." It seems to be a software-specific scenario where a change is proposed, users say, "I don't like it", and then engineers get somehow upset about that fact, rather than just saying, "cool, it was an experiment. Let's revert and move on."
I did not get into programming because I love computer interfaces. For me, being treated like a professional means that a company makes me specifically more productive. I don't care if a change makes things better for someone else if that change is making it harder for me to do something I love. As a professional developer, it is OK to demand tools that work well for you. Again, this is the case for every other field -- no one expects a professional woodworker to switch to a new measuring system that they don't want to use, even if that system is popular with some people.
Of course no field gets this perfect, but virtually everyone gets it better than us. I'm not demanding a theoretical utopia I can only imagine. I'm looking at every other professional field and saying, "why can't we have the stuff that they have right now?" It's very much a feeling that's informed by how good it feels today to work with physical mediums as an artist, and how utterly crappy it feels by comparison to use Skype or Windows 10.
And if that means that software moves slower, who cares? It's not theoretical to me. Other mediums are better to work in -- so whatever they're doing, we should copy, because they're better than us.
Except for the changes like "reduce memory usage" or "fix crash when X happens" --- those are true improvements.
To expand on your point, how many bugs and crashes do we tolerate in software simply because rebuilding features has priority over refining features or handling more edge-cases?
Slack's new interface isn't just different, it's buggier. Stuff like copy-paste is broken. So it's not just that we have to adapt to a new workflow, we're also accepting a lower-quality piece of software that is less reliable than what we had before. And in theory a refactor or rewrite might be so valuable that I could tolerate that, but I don't see that value here.
Speaking of muscle memory, I still remember these … and — from the time when most web apps understood HTML. Too bad they stopped…
This exchange is a pretty good summation of one of the biggest purely practical reasons why I'm so obsessive about tools, and why I'm so willing to put up with the initial cost of learning systems like Linux and Vim/Emacs.
Outside of fundamentally better workflow improvements, most professional fields don't randomly change their tools. If you gave a professional artist a new pencil that had to be gripped differently for no reason, they'd throw it in the trash.
But in software, we tolerate buggy tools that change all the time for no discernible reason. We tolerate software that simultaneously targets professionals and casual users, serving both segments poorly. We tolerate software that can't be customized or adapted for specific workflows. It's tough to put into words, but if you watch a musician or a painter interact with their tools, there's a very clear difference that emerges, and over time you start to realize how much better all of their stuff is.
In most professional artistic settings, workflow changes only happen because they have a clear benefit -- drawing from your shoulder instead of your wrist, changing your embouchure if you play an instrument. And even in those fields, it's generally accepted that over time people will end up with very specialized setups that are very consistent and refined and that remain constant for years and years.
Only in the software industry would someone tell me that my professional tools should change because change is inherently good. Only in commercial software would an elegant, consistent interface like Markdown that allowed me to build up decades of muscle memory until my computer was an extension of my fingers and I didn't need to think about the way I typed -- only in software would that be considered a bad thing.