[rich_sasha]

One of the most expensive pieces of software around is a Bloomberg Terminal, with base price starting at 20k/yr/license, plus more for extra data. And the UI, when you first use it, is beyond clunky. It looks seriously like a piece of stone age technology, adapted from clay tablets. Even the styling is like 80s films about edgy hackers and their punch cards.

Except when you talk to the grey hairs, you realize that UI has never, ever changed - it is backwards compatible back to, well, stone age of computers. It is quirky, but once you learn it, you're done for life - no refreshed new looks or skins or dark modes (well, it's always in dark mode) or rewrites in Svelte or whatever. That's basically because the power user is essentially the only user that matters. They know all the arcane key bindings and weird abbreviations, and Bloomberg knows better than to mess with it.

I hated it at first but grew to love the stability of it.

[dchest]

Speaking as someone who has never used it but has spent some time researching it, the Bloomberg Terminal constantly undergoes UI changes, though not in a dramatic way. It's obvious if you look at screenshots throughout the time (it even had some gradients!). It has had its own "rewrites in Svelte", transitioning from a custom renderer to HTML/JavaScript.

But you're correct - they don't mess with it, they slightly and mostly invisibly improve it, and someone who learned it in 80s could use it without problems today.

https://www.youtube.com/watch?v=uqehwCWKVVw

https://www.bloomberg.com/ux/2017/11/10/relaunching-launchpa...

https://www.bloomberg.com/company/stories/how-bloomberg-term...

https://www.bloomberg.com/company/stories/designing-the-term...

[asoneth]

> the Bloomberg Terminal constantly undergoes UI changes, though not in a dramatic way

This is correct. (Source: I worked on Bloomberg's UI change management policies.)

Despite dismissive comments from design industry folks and more modern-looking competitors, the folks who ran Bloomberg's UX team maintained a focus on customer needs and user research. There are even a few cases where function teams went back and re-implemented old "bugs" after a rewrite (e.g. in the MSG editor) because users had adapted to the old behavior. (Thankfully nothing as bad as spacebar heating https://xkcd.com/1172 though)

[SecretDreams]

> and someone who learned it in 80s could use it without problems today.

That's the true dream. Like all of those old movies where the hacker or fighter pilot has to use some foreign, alien or futuristic tech and they just use it!

[MetaWhirledPeas]

I just watched a few minutes of someone using it, and I have a few observations.

- Uniform menus everywhere

- Every menu has an ID visible in the corner. Imagine how easy troubleshooting would be if you could just say, I'm on menu 4388 and I'm not getting the result I expect.

- Every selection has a number. Presumably you can type this in rather than mouse over to it?

- Every page has keywords you can string together for instant searching

- No transition animations

This is nice to see: a tool for getting things done, not for nudging the user in a desired direction to satisfy marketing goals.

[myth2018]

My parents work for a company using an old system using numeric IDs for menus and screens. It's currently a web app, but it seems to share a good deal of code (and a great deal of philosophy) with the previous, older version, a TUI apparently built in Pascal.

I can confirm that those numeric IDs help A LOT in troubleshooting and documentation. And not only that: those IDs are frequently and naturally used by the users themselves when communicating between them. Needless to say, they also use the IDs to navigate the system, only touching the menus when they have to use some infrequent function.

I always mention this "case study" to UX folks who insist to dumb users down with childish interfaces.

[asoneth]

> no ... rewrites in Svelte or whatever

The vast majority of the terminal interface has been rewritten more than once, but the UX framework team does a great job mimicking the prior look and feel each time. Though if you know what to look for you can spot functions that haven't been updated in a while, and even a handful that have remained unchanged since the beginning.

> no refreshed new looks or skins

Basically right, though Launchpad was completely new and PDFU COLORS <GO> provides color themes intended for folks with color-vision deficiency.

[gedy]

Yeah this is a good reminder that technology rewrites do NOT need to rethink the UI. I've always favored updating the technology first to enable faster incremental UI changes afterwards.

I love my UX friends but they almost universally hop on tech updates to rethink everything, and then it muddies up the customer feedback on the rewrite. Was this a tech bug introduced? Or an annoying UX change? etc.

[asoneth]

When I taught UX graduate students we certainly tried to instill in junior designers that their chief responsibility is to their end-users. And where it doesn't negatively influence their users their secondary responsibility to the business. Coming it at a distant third is their own sensibilities and satisfaction. Putting business needs above your users makes you a black-hat designer peddling dark patterns. Putting personal sensibilities over either makes you a digital artist -- not inherently a bad thing but also not what most companies are looking for.

Frankly many junior developers seem to have missed this lesson as well. I've seen countless rewrites for personal satisfaction or learning rather than concrete end-user or business goals.

[throwaway24124]

Well, it depends on the audience of your software. Bloomberg Terminal is awesome, but has a steep learning curve. Most people who learn to use it are learning it with the specific intention of finding a job where they are paid to know how to use the Bloomberg Terminal. I agree, the UI is awesome for that, but if your startup is an app to find a dog walker or something, the vast majority of your potential users are going to be turned off by the Bloomberg Terminal interface.

[edelhans]

Yeah but even if your app is the new Uber for dogs the goal should still be to let users accomplish their task as fast as possible and not some app engagement metric.

[eviks]

> you realize that UI has never, ever changed - it is backwards compatible back to, well, stone age of computers. It is quirky, but once you learn it, you're done for life

True power users could customize to remove the quirks and also be set for life, but at a better level of ergonomics. Or they could even use the best customization from one of those gray beards who cared about ergonomics more ever back in the day

And none of of the cheap criticisms of skins would change it since skins for power users are optional

[dacryn]

You fail to grasp the value of bloomberg terminal.

The UI has in fact, evolved, but it has never changed. For example, higher DPI screen sizes, the UI is now instrumented in a web browser, no longer the the old TUI. It is fast, it is familiar, it's the same, but it evolves, if that makes sense.

If you know how to use it in company A in decade 1980, you know how to use it in company B today. That doesn't mean it hasn't improved or improved ergonomics.

It's a beautiful piece of engineering that got the basics right. Power users add whatever they need to it, modular, but it's not like Vim or VSCode where you are basically useless without a large effort when moving into a blank new updated version, let alone things like the ribbon design vs the old design in office.

[eviks]

It's the other way around, the value of that terminal is in the information, not whatever hated UI quirks it had been stuck with since its inception, yet people keep falling for that old logic "old+expensive = great".

> it's the same, but it evolves, if that makes sense.

it doesn't, these are the opposite.

> If you know how to use it in company A in decade 1980, you know how to use it in company B today. That doesn't mean it hasn't improved or improved ergonomics.

Neither does this: it would be just as trivial to select at company B "use config ergonomic_grey_beard_1980" and continue with all your knowledge, just without those quirks you hated in 1980 that led you to change the stable defaults to a better config.

> but it's not like Vim

And in some sense in the relevant UI area it's exactly like Vim, where many bad quirks in the default config are praised by the grey beards and new converts alike.

> moving into a blank new updated version

Why would you do that instead of using your old config???

> the ribbon design vs the old design

Neither is forcing a change like this the only alternative

[Suppafly]

I worked for an insurance company for a while that was pushing their agents away from 'green screen' terminals to webapps that did most of the same stuff, the agents hated it because they had every step necessary to do quotes and such memorized and were way slower trying to navigate menus on the web.

[nextts]

I think to me that would be the appeal of Vim. VSCode implicitly promises this with its plugin approach. Using Ctrl Shift P for years is the sort of thing I like. One less thing in my way.

[masfuerte]

I'd never noticed that ctrl-shift-p was a thing. In gvim ctrl-p moves up a line but ctrl-shift-p seems to do page up. Do you know if this is documented anywhere?

[msy]

It's not just about 'UX backward compatibility', it's the epitome of an expert user interface. If you don't know what you're doing it's near unusable but once you've mastered it the speed at which you can operate is incomparable to 'normal' software. It's also one of the few examples of this outside of software engineering tools.

[mfld]

> well, it's always in dark mode

So it's no TUI app that can accommodate changes to the terminal emulators color profile? I am a bit disappointed.

[intrasight]

> backward compatible to the Stone Age

Like Emacs. And myself ;)