I have the highest regard for these terse, but information-dense displays that enable humans to quickly grasp all information they need to do their jobs, and nothing more. This is also something avionics provide - a modern cockpit displays are a work of art in layering needed information for what is happening at the time, both the dense PFD and the much specific that communicate aircraft status.
I know it's impossible, but I've always loved to get a hands-on demo... I'd love to learn the UI patterns involved, for example. I can always make my Emacs faster to interact with, dammit!
I had occasional access to a Bloomberg terminal 20 years ago and I still miss it. I miss it a lot. I wish they offered a cheaper option for users who only need to use it infrequently. I mean, $25,000 annually is brutal if you're not a broker-dealer.
What does the terminal really give you that you can't get from various other data sources scattered around the Internet, like your brokerage's own API, third-party data sources, various forums and subreddits, ...? Is it just the real-time firehose nature of the thing?
"Before it's here, it's on the Bloomberg terminal".
Next week it will be in Bloomberg Business Week, long after it had trading value.
Bloomberg has its own news collection and analysis operation, bigger than almost all US newspapers.
For many years ending around, a Bloomberg BusinessWeek paper magazine subscription was an incredible bargain. It was often possible (through tons of discount magazine sites) to get it for like $25 a year or less, and you could link that subscription online to get access to a lot of Bloomberg.com articles. Far from Terminal access of course, but I still really enjoyed the in-depth journalism and analysis. Around 2020 or so they pulled it off that type of distribution, now it's $100 a year and doesn't come with any online access other than the magazine articles themselves. And the full subscription to digital is $400 a year. It's priced just high enough to make it hard to justify for those who aren't real finance professionals. I really miss Bloomberg since the hard paywall went up.
As for the $400 a year, for years I'd cancel before the end of the subscription, ignore them for 2 weeks, and get a great offer for the next year. I haven't done it this year, though, forgot to set a reminder.
Great value anyway. One of the only news sources focused on facts and meaningful stuff. Its been my go to news source for at least 15 years. I don't work in finance.
That's the default. You can get real-time feeds for pretty much anything if you pay extra for it. That also really adds up BTW so only our traders tend to be permissioned for real time data.
> Imagine that your ISP (connectivity), browser (interface), OS (orchestration), app store (functionality), search engine (directory), social network (chat), and media houses (news & data) were all run and owned by a single company.
That's called vertical integration, not an anti-trust violation.
It could be if they were the only one trying to offer this, but because it's so expensive, there are always multiple competitors trying to pick off their users. And when the next round of corporate cost cutting comes, they always come for the Bloomberg terminals, trying to push something cheaper.
But the cheaper alternatives are always inferior too (never mind the decades of muscle memory on Bloomberg).
Despite the price Bloomberg has maintained its number one position. I think in no small part that's also because their customer support is absolutely excellent. Press F1 twice and you're in a chat with the helpdesk; these days the initial query is triaged by AI of course, but for actual issues you're talking to a real human in less than a minute (sometimes just seconds), and these real actual humans typically have the expertise to help you. If not, the higher levels of support are good too. Don't underestimate this aspect!
Yeah, my reply’s a bit tongue-in-cheek. I don’t really oppose monopolies per se – it’s a free market, a better product could compete and win.
But one of the Bloomberg’s main selling points, at least from the last discussion I’ve read on HN (and some of the comments here too), is the chat, which means there’s network effects at play. So you can’t just build a better terminal, you also have to convince the majority to switch.
(Though apparently Bloomberg Chat also supports email? Maybe the competitor can use that to interface with it, so it’s only “green bubble” bad, not “all your colleagues are there and you’re not” bad.)
Ah, I see and didn't pick up on the tongue-in-cheekness!
The network effect is real and I don't think you can interface external email with it. You do get a Bloomberg email address as a user but I've never seen anyone in a chat with an external email only.
Maybe it's because of all the integration that's hard to do. A bit like if Windows 98 was inseparable from Internet Explorer, so you couldn't just replace one or the other.
- an email address (firstname.lastname@bloomberg.net) used to email and chat
with other professionals on the Bloomberg network. This is like LinkedIn and
WhatsApp combined for people in finance, and people can keep the address for
two years after e.g. losing a job, so they can use the network to find a new
role.
- analytical software: tools to calculate various KPIs to assess and analyze business opportunities.
- capability to transact: The BB terminal is not just an information system, it also permits you to buy and sell directly.
(My university has one BB terminal that MBA students can access and get training on.)
It's all that and much more. It has an enormous amount of data of all kinds that could be vaguely relevant in finance. BMAP can show you vessels (AIS data), location of LNG terminals, powerplants, bank branches, crop yield forecasts (just a few I picked off the list just now) and so much more on an interactive world map.
And every kind of price you can think of (almost). Crack spreads? River levels? Baseload power? Airline schedules? I am trying to think of more but it's truly VAST.
They have custom screens, configurable screens, custom settings (you want the generic futures contract to be rolled differently from the default? Sure...), option pricers and stuff I have never had to search for or look at.
They also have their own query language BQL for data queries and the BQNT platform is Python-based and optimized for use with the data.
If you just want the latest stock prices for your portfolio your online brokerage app probably does a better job.
The biggest miss right now is their ASKB AI chatbot, which is borderline moronic. It has to be some older model with low thinking/effort, because the first answer it gives is always broken in some way. Hallucinated functions or fields, outright wrong descriptions, etc. I don't even try to use the first answer anymore but ask it again to check its previous answer against verified and authoritative sources/documentation and will come up with many corrections. And then I do it again, and maybe a fourth time.
> I don't even try to use the first answer anymore but ask it again to check its previous answer against verified and authoritative sources/documentation and will come up with many corrections. And then I do it again, and maybe a fourth time.
Maybe all they need to show massive improvement is a for loop of four iterations with the loop body being "are you sure, check the docs?" and return the result at the end.
From what I have seen of it - for the people who can afford it I suspect that having everything all in one place under a common UI is a bit of an enabler.
Reducing the friction or feedback loop of information retrieval/analysis can change how people work just like repls and hotreloading did vs slow and long compiles
I've heard many people quip that the best thing about using a $25000 terminal is being connected to other people who can afford spending $25000 on a terminal.
Its breadth and depth - it’s about being able to see a forward for ISONE Vermont Pk power prices, swaptions on uk gilts, where 5yr forwards are pricing for the yen, freight rates and ship tracking, weather data, satellite analysis, legal and legislative data etc.
You hit HELP HELP (a dedicated button on the keyboard) and a real, knowledgeable person in your time zone will _fix_ any problem with the data you see on your screen. If it's not fixed by the time their day ends, they will hand your ticket to yet another knowledgable person to the west who will continue working on it to fix it. That is, the data quality, and the data quality processes, are a large part of what you pay for. People get hung up about real time, but what if a number on a financial statement from 10 years ago is screwing up your model training?
Some markets, for example, corporate bonds, are mostly on BB chat. They are dealer markets. And dealers will provide “runs” of prices on chat. List of 200 bonds and their prices.
So if you want to trade bonds, need BB as that is where it is happening. Keep in mind some of these bonds trade infrequently (sometimes few times a month).
Having said that, there are more and more bonds trading electronically all the time.
Yeah, I think this is something a lot folks miss when doing UI/UX these days. There’s an immense focus on new users, not on folks who rely on it day in and day out for actual work. I’ve seen loads of old, incredibly “ugly” UIs that absolutely mog newer friendlier attempts at the same. They’re absolutely intimidating to new users, but once folks learn it they’re indispensable and obviously intentional.
I really wish there was a resurgence of information dense UIs and tools, optimizing for people who use them daily and not “use it once.” It really reminds me of the hate folks have for things like Eclipse or IntelliJ which are daunting at first but once you get over the shock and learn them really are amazing tools.
You’re reminding me of this thread I found a long time ago where someone was talking about the old amber-screen library computers that felt FAST. The gif here shows what it would look like to read Twitter in that manner… https://x.com/apaprocki/status/1230241842928791552
The difference is that Costco pays their employees well enough to keep them around (notice that Costco prints the start year on each employee badge- average retail length of service is 1 year, Costco keeps people around for decades) and so they can learn to use this older software and master it. Most retail operations optimize their inventory management for new people, because they don't expect someone to stick around long enough to master it.
My wife worked at a small regional chain of pharmacies that had an ancient MS-DOS based EHR system. And she mastered that system to the point where people would call her when she was home with me and ask her how to do something, and she would walk them through "okay, hit tab three times then shift-F12 then enter the frabble and then alt-F3 to go to the next screen and enter the drabble." It was difficult to learn it, but if you mastered it you could do everything you needed to do from just muscle memory while you were on the phone arguing with the doctor's office that their prescription was not correct and needed to be changed.
Then the small chain got bought by a national player and their brand new GUI system was brought in. It was a lot easier to learn- discoverability was a lot higher with menus and drop-downs and all of that. But if you stuck around to master it, you could never be as fast as you could with the older system. And that emphasis on quickly integrating new employees- but lower payoff for experience in employees- was clear in every choice that national chain made. They wanted their employees to be completely interchangeable, which meant that they didn't care if you had six months of experience or four years of experience or three decades of experience.
Same story at my bank. Tellers used to speed run through a TUI, hitting tab and other key combos to accomplish things. Now it's a long wait with them mousing around.
The best thing about all character-based applications I've ever used is the keyboard buffer. Even if you're using a TUI on a system that has a slow backend, it almost always blocked the thread and didn't consume from the buffer during waits, so if you knew the sequence of keys was "Total, F1, 1, Enter, Enter, Tab, 2125551234, F10, 1, Enter" you could just enter all those keystrokes as fast as you could and the system would eventually catch up with you, meanwhile you were bagging a purchase or counting out the cash, whatever.
The GUI also has lag - especially in the era of web-technologies-bloat, more lag than ever - but whether touchscreen or mouse there can't be a buffer, so each lag requires you to stare waiting for the next UI to be drawn.
Unfortunately web-inspired TUI toolkits start to lose that property. Hit a key too fast, before the next screen runs code to attach its event listener? Sorry your key is lost.
Clickable web GUIs that don't lose your interactions are even worse. The email client I have to use at work buffers your "mark as read" click when it's loading new letters and applies it to the one that's under your cursor when it finishes loading them.
This is a good point: you can build character-oriented interfaces "for dummies" — IBM i née OS/400 which Costco uses is actually good at this, with a standard interaction design centered around menus and form interfaces that map directly to CLI commands — and information-dense GUIs, but you can't very well build point-and-click interface with keyboard-style buffering.
Agreed. Also, you can try to satisfy both new and expert users. It take more thought and design, but that principle is in both Nielsen's Heuristics and Shneiderman's Golden Rules.
> Most retail operations optimize their inventory management for new people, because they don't expect someone to stick around long enough to master it.
Somehow I think I failed to think very deeply about this very important link! The twin happenings of the elimination of any retail career that pays a living wage, and the availability of GUIs and especially touch-screen GUIs happened simultaneously over a couple of decades, and the touch-screen dummy interface naturally won out. Not just because training that used to take two days for the 'green-screen' system I learned in 2001 now takes 15 minutes of instruction (and even that training is rather redundant for anyone over 95 IQ) -- also because the massive differential in productivity of an expert vs. new cashier/teller/pharmacy tech/etc. only even manifests if people are on the job long enough to become expert. And they're really not anymore.
Kinda why I prefer self-checkout. Even though the UI is firmly in "dummy" territory and I have to use a touch screen, I've used the current iteration of the self-checkout software more transactions than most cashiers I meet at Target, since they seem to have been there about 2 weeks on average.
The de-skilling of retail clerk I think predates even the GUI. I think it goes back to the laser/upc/database systems of the 1970's and 1980's.
In 1970 retail clerk is a medium-skill medium prestige job. There were lots of stories of people who started as a retail clerk and end up as CEO, and even people who aren't that ambitious could still carve out a comfortable middle class lifestyle as a retail clerk. But that's because there was a reward for experience. Before the laser/UPC/DB system rolled out, you needed at a glance to be able to know the price of an orange, or the difference between the expensive and the cheap mushroom. Because there would be a paper tag on every cheap mushroom that said "10 cents", but what if someone swapped that paper tag onto the expensive mushrooms that were supposed to cost 30 cents? You needed to know the difference, and have the real prices memorized to be able to be fast enough to keep the customers happy.
This also meant that you couldn't have a large retail store (something at the scale of the modern Walmart or Target), because you couldn't get enough clerks who could remember all the necessary prices and so inevitably you would lose money to tag switching. Department stores solved this by forcing you to buy each good separately in each department- you would pay for your housewares in the housewares section and your clothes in the clothes section, and they had another whole army of clerks in the basement managing store credit to keep it low-friction enough. But that's very high-touch, demanding a lot of workers, and those people still have the rewards of experience so they are all comfortable careers.
But then the laser/upc/database system came out. And suddenly you would get a natural language description of the item (or even a picture in the modern systems) from scanning the UPC. And you could scale the store up as much as you wanted, just needed to buy a bigger hard disk for the database, so the stores tilted more towards capital and less towards labor. And that eviscerated the career path of most retail stores, because store owners no longer believed that keeping retail clerks around benefited them.
This is machines hallowing out a job and alienating workers from their labor for the benefit of capital- exactly what Marx had described a century earlier for factory workers in Das Kapital. But for reasons of class and gender we mostly ignore retail clerks so it doesn't seem to have attracted near as much attention.
I think that's the 'push/pull' problem though. Information dense UI that takes months or years to learn, or something that may be less information dense, but a monkey can learn it in minutes.
For me, it's typically a balance between speed and efficency of data to help you make decisions quickly, but should be able to be used by anyone with technical experience.
So less of both. Less 'pretty UI' and less "omfg here is every piece of information shoved together and you need to be a wizard to understand it"
The same for hardware. Where I used to live, the ticket vending machines had you punch in your 3-digit station code, adult/child, single/return, and boom ticket issued. You had like 6 button pushes and could buy a ticket in seconds.
In this video I made the mistake of using a debit card (still very slow to clear in 2005) which adds the delay https://youtu.be/pQCktMWR94Q?t=7
They replaced them with Windows-based touchscreen vending machines where each screen selection took as long as the whole former selection process took just to load.
I love displays from space vehicles and nuclear plants for the same reason. Data and controls organized in a dense but presumably highly usable manner. And it covers the more specific view of an expert in a particular subsystem, as well as the higher level view of those responsible for the whole system.
I remember seeing glances of the Mission Control screens during the Apollo and early shuttle program - two screens and two channel selectors where the programming would be a screenful of real-time data rendered in the adjacent RTCC (real time compute center, or something like that) and captured by slow-scan CCTV cameras and piped to the consoles as TV signals.
UI designers who were trained in the currently-popular paradigm would be at a complete loss to design such things. "Swipe left from the home screen, tap the nothingness on the screen to reveal the controls overlay, find the ••• in one corner of the screen, tap "More" in the 3-item menu, and then tap 'Bond Quote' in the modal dialog that takes over the entire screen." Keyboard shortcuts? "Hmm... interesting idea. Never thought of that. We might add those in a V2."
I got a rental the other day and while driving could not find the clock! I had to hit TWO BUTTONS on the infotainment touchscreen panel to get back to "home" and see the gd time. I was driving on curvy roads in Yosemite too. Truly abhorrent.
Also bad: When a software update moves essential elements around.
For the last three years I've looked at a particular location in my car for the clock. Then a software update moved it not just to a different location on the screen, but to an entirely different screen.
Three years of muscle memory is probably going to take another three years to un-do.
(For those of you who don't understand why a clock is an essential part of driving, it's so I know if I can park in a 9a-4p zone, or if it's the time of day when I can use the reversible lanes in my direction of travel, or if I can use the lanes that are HOV sometimes and HOT at other times, or a hundred other things.)
The same software update moved the wiper locking controls (for car washes) from a button on the left steering column stalk to a menu option four levels deep on the center screen.
That's definitely surprising in this age. Could you change the air conditioner ui to celsius? But changing to kph is not something I expect by default in _every_ car. I do expect a visible clock in every car though.
The corollary of which is that no function the driver might sensibly need while underway should be presented or accessed through a touch screen.
So physical levers and buttons. With some tactile feedback, such as raised lips or centre bumps or distinct shapes, so without looking the user knows which one their finger is on (third from left in the Aircon group). Displays can be present multiple items (don't need physical dials) but changing that display when driving comes back to buttons.
My car has user interface elements that you would want to use while driving (e.g. to cancel the navigation guidance) that only become visible if you hold your hand in front of the screen....
My car display has NO WAY to cancel navigation guidance from the GPS screen. You can pause guidance, but when you turn the car on again, it tries to resume.
You have to go back from the guidance screen to cancel. So, I never use it.
> But it’s not like the designers are unaware of more information dense alternatives.
Strong disagree. I have been in UX reviews for products like Bloomberg terminals as the intended user. I specifically asked for information density and the designer looked at me like I was from the moon. Their counter offer was a permanent floating chatbot that covered the small amount of information still on the page. No joke. Information density is something those web designers genuinely could not comprehend.
“ You can remove the drawn-in borders just by adding a little visual space, and use subtle color differences to separate rows if needed” ———> “Let’s make it 10 rows per screen (and somehow still with low-legibility font choices and low-readability text design)! Make the design breathe, you know, white space!” ———> Me, having an aneurysm
I described it as being lost in an ocean of white pixels. Then they added padding and broke hyperlinks. I know it was incompetence but it started to feel malicious.
I would love to read up on how such interfaces (spaceship flight controls, avionics controls, etc.) are designed and the history of them. If anyone knows of any such resource, please point the way!
This is the difference between accessible and efficient.
Bloomberg, and aircraft controls are expert interfaces, they have steep learning curves and in the latter require case extensive education to use (Bloomberg also does a ton of work on user training). Which is absolutely fine when efficiency trumps everything else and it's a system people use all day every day.
Most interfaces we use on a day to day basis are the opposite, they're designed to be accessible (in the sense that you can use it if you are unfamiliar, not disability) for someone unfamiliar, untrained, and uninterested in learning, which also aligns a lot of the time with sales growth. Nobody is willing to do a 6 week course to use the new HR UI really, really efficiently. The unfortunate thing is some of these are enterprise UIs some of us do have to use all day every day and that's where the impedance mismatch gets painful.
> Most interfaces we use on a day to day basis are the opposite, they're designed to be accessible
I read this and then I think of glass doors with handles on both sides that I don't open till I read the instruction on the affixed plate: PUSH/PULL. The door forces this decision on every single conscientious user. These things are everywhere, including software interfaces.
Accessibility peaked around Windows 95 and Windows 98 when buttons were clearly buttons, scrollbars were clearly scrollbars, and clickable text was in blue and underlined. We're now well past that, and without familiarity with disappearing scrollbars, three-dot menus and other obscure means of navigation, it's very unintuitive to get around. We don't even have manuals anymore if you do want to learn how things are supposed to be used.
Strongly agree with you. Doug Engelbart (of mouse + groupware + hypertext fame) once said that if ease of use were the only design criteria, we would all ride tricycles instead of bicycles.
Not every user interface has to be or should be walk up and use. But there also needs to be a clear payoff to training.
> Most interfaces we use on a day to day basis are the opposite, they're designed to be accessible (in the sense that you can use it if you are unfamiliar, not disability) for someone unfamiliar, untrained, and uninterested in learning, which also aligns a lot of the time with sales growth.
Must be why my fghxjing iOS jorard doesn’t work, and why searching for “preferences” (which is what it is called on Mac) doesn’t open “Settings”. Bonus points for gazillion of gestures that require the Necronomicon to decipher.
The modern Terminal is based on a private fork of Chromium to give it the look and feel of a VT100 terminal, and integrate their private networking and security technologies. The BBT predates HTTP, and backwards compatibility is hugely important to the company- they have a museum where a second-generation Terminal from ~1985 shows the current news, because they are so dedicated to backwards compatibility that they still can support the 1985 hardware.
This isn’t really true. The old terminal monitors are essentially dumb CRTs and the the museum pieces are just displaying the output from a regular PC running the modern app converted to an appropriate signal + BNC connectors so you see something and not just a black screen.
The old terminal hardware communicated over dedicated serial lines and worked in conjunction with many other “server-side” pieces of software that no longer exist, so it would be impossible to make one function as it did back in the day.
I would love to learn the specs of the original 1983 terminal. By the look of it, it's NTSC timings, so 200-ish lines, but it shows text and graphics, so it must have some bitmap capability. If I were to replicate the look, I'd use something that can display ReGIS or NAPLPS.
They ran iRMX on an 8080 variant and had graphics card that handled the A/V aspects. I suspected the displays could have actually been PAL because it had slightly higher resolution, but I haven’t been able to confirm that yet. I have one of those “Smithsonian” pieces on my desk waiting for me to have to some free time to try to repair the CRTs when I someday have more free time than I do now (any good resources for that??).
Hey Andrew, former BB employee here. What's the dev experience like over there now? Back in 2011 when I left, RnD was leaning heavily on the custom JavaScript + IDE for frontend dev, and mostly cpp and non-relational mmap DBs on the backend.
How have things evolved?
So much has changed since that time. There's a great recent talk by Jason Williams and Paul Williams that goes into a lot of detail [1]. Most things now use TypeScript developed natively in VS Code using extensions and a CLI for the additional proprietary bits. That all nicely integrates with agentic workflows, skills, etc. ComDb2 is very relational, very battle-as-in-jepsen-tested and has been open-source for a long time now [2]. C++ is still king, but there's plenty of Python, Go, Rust, Haskell, OCaml, etc. running around.
Ok, lemme ask you this: What has been your favorite Bloomberg keyboard over the years? I preferred the mid-2000s mechanical keyboard before they started going lower profile keys, which suffered from off-center binding when pressing near the edge of a key.
I got my eyes checked a few years ago and they were literally using a VT100 interface. The person knew the keyboard interface like the back of his hand and it was so much faster than a mouse-based GUI that is so prevalent.
Bloomberg's interface looks overwhelming at first, but a trader who knows where everything is can scan a lot of information very quickly. There's probably a useful distinction between an interface being diffucult to learn and an interface being inefficient to use once you have learned it.
Starting October 14, 2026, Bloomberg is requiring a physical Bloomberg Keyboard for login and access on all Bloomberg Open Terminal workstations.
The harder Bloomberg tries to control and restrict its users, the more users will just find alternatives or even build their own proprietary systems. It thinks it can be Apple in the financial data world, but I highly doubt it.
I doubt that's a big blocker - the keyboard is one of the selling points and I'm guessing that you could just have 2 keyboards if you really wanted to (the Bloomberg one for auth and the other one for longform typing??)
Maybe related, but is anyone familiar with a series of trading terminals from BT in the UK? I cut my teeth in circuitry repair working on these models like the 405 and 405i in the early 2000s.
These days I can't find hide nor hair that these things ever existed!
Any time there's an article that mentions the Bloomberg terminal I always have dig around for this mythical series of trading platforms, alas I now worry my memory is playing tricks on me.
I wish that article showed what the display looks like instead of just the keyboard. Others here have mentioned the lovely information density; I'll have to search elsewhere to see it
The article has an image of terminal keyboard used by a legendary trader... the key-caps are immaculate; the letters haven't faded at all. Amazing. I wonder if they were double-shot...
On one hand, at the price point of >$10000 for the whole terminal, it's quite possible that >$100 went into the keyboard.
However, I suspect you're coming at this from the perspective of a coder or a gamer or some other category who types thousands of keystrokes per day (tens of thousands?). With the assortment of special-function keys that enabled the user to execute common tasks with only two or three keystrokes, I suspect that even actively-used Bloomberg terminals may see an order of magnitude less keystrokes.
(On the third hand... I seem to remember reading that the Bloomberg system included chatroom functionality, which may increase keystroke count)
> I especially love that Gross taped his login and password directly on his keyboard, which makes the object more relatable. I may never know what it’s like to manage billions in assets from a Bloomberg Terminal, but I absolutely understand the trial of remembering my passwords.
Wow, what a great history artifact. It is hard to look at this and not wonder where did we go wrong. Everybody _should_ know better, and yet here we are. And now immortalized in the National Museum of American History for all posterity.
From memory when I worked there 10 years ago, all the Javascript based stuff was pretty slow in a very noticeable way. They've likely gotten better as they are using a well known Javascript and rendering engine, so they're getting many optimizations for free.
We’re in the middle of TC39 plenary in Tokyo right now as a matter of fact! Rob regularly posts on X as things advance through the staging process, and today is the final day.
Love how those dense screens make numbers checkable at a glance. Tiny shops still live that gap—bank lumps vs invoice lines—just without a dedicated terminal.
The current models are just USB keyboards with special buttons. The requirement of an official keyboard will increase the second-hand market prices, but they used to be available on eBay.
In finance, people who consider the cost of a Bloomberg terminal "expensive" are often considered "not serious" (small fish)... it has often been seen as a status symbol as well as a professional work tool.
Not really. If you ran a restaurant instead, you may also have to buy a few thousand dollars worth of ingredients and pay thousands of dollars in salaries per year and renting a venue to make the revenue that you are making. Even a pizza delivery service may need a few thousand dollars a year for a car and its driver plus petrol, taxes, insurance. Relevant is not the cost of doing business, but the annual revenue and margins.
E.g. $32k cost of business to make $2bn revenue statistically approximates zero.
Not to sound like a YC fanboy here, but I love how pretty much every great business can be traced back to a start which somehow follows what YC talks about today.
I guess the first principles of what makes great Founders and great businesses from back in the gilded age to today have stayed the same.
I have the highest regard for these terse, but information-dense displays that enable humans to quickly grasp all information they need to do their jobs, and nothing more. This is also something avionics provide - a modern cockpit displays are a work of art in layering needed information for what is happening at the time, both the dense PFD and the much specific that communicate aircraft status.
I know it's impossible, but I've always loved to get a hands-on demo... I'd love to learn the UI patterns involved, for example. I can always make my Emacs faster to interact with, dammit!
If you're in New York City, the New York Public Library has a terminal you can use.
Wow, really? Do they limit your time on it? I would imagine that thing is in constant demand.
It’s a two hour time limit and you can only take pen and paper with you inside (no photos/data export)
There were two in the Greenwich CT library when I lived there in 2019, not sure if they’re still around.
> There were two in the Greenwich CT library
Of course. That's where all the hedge fund managers live.
There was one for visitor use at the Bank of America branch on Sand Hill Road in Silicon Valley at one time. But BofA closed that branch.
Catching up on the news while the kiddos do group storytime?
I had occasional access to a Bloomberg terminal 20 years ago and I still miss it. I miss it a lot. I wish they offered a cheaper option for users who only need to use it infrequently. I mean, $25,000 annually is brutal if you're not a broker-dealer.
What does the terminal really give you that you can't get from various other data sources scattered around the Internet, like your brokerage's own API, third-party data sources, various forums and subreddits, ...? Is it just the real-time firehose nature of the thing?
Structured, deep and wide
The big one is that almost all of those data sources are subject to a 15 minute delay...
"Before it's here, it's on the Bloomberg terminal".
Next week it will be in Bloomberg Business Week, long after it had trading value. Bloomberg has its own news collection and analysis operation, bigger than almost all US newspapers.
For many years ending around, a Bloomberg BusinessWeek paper magazine subscription was an incredible bargain. It was often possible (through tons of discount magazine sites) to get it for like $25 a year or less, and you could link that subscription online to get access to a lot of Bloomberg.com articles. Far from Terminal access of course, but I still really enjoyed the in-depth journalism and analysis. Around 2020 or so they pulled it off that type of distribution, now it's $100 a year and doesn't come with any online access other than the magazine articles themselves. And the full subscription to digital is $400 a year. It's priced just high enough to make it hard to justify for those who aren't real finance professionals. I really miss Bloomberg since the hard paywall went up.
As for the $400 a year, for years I'd cancel before the end of the subscription, ignore them for 2 weeks, and get a great offer for the next year. I haven't done it this year, though, forgot to set a reminder.
Great value anyway. One of the only news sources focused on facts and meaningful stuff. Its been my go to news source for at least 15 years. I don't work in finance.
That's the default. You can get real-time feeds for pretty much anything if you pay extra for it. That also really adds up BTW so only our traders tend to be permissioned for real time data.
You get the chat: https://news.ycombinator.com/item?id=49286955
this answers the question pretty comprehensively i think. https://theterminalist.substack.com/p/bloombergs-7-powers-an...
> Imagine that your ISP (connectivity), browser (interface), OS (orchestration), app store (functionality), search engine (directory), social network (chat), and media houses (news & data) were all run and owned by a single company.
Oh, cool! An antitrust violation epitome!
That's called vertical integration, not an anti-trust violation.
It could be if they were the only one trying to offer this, but because it's so expensive, there are always multiple competitors trying to pick off their users. And when the next round of corporate cost cutting comes, they always come for the Bloomberg terminals, trying to push something cheaper.
But the cheaper alternatives are always inferior too (never mind the decades of muscle memory on Bloomberg).
Despite the price Bloomberg has maintained its number one position. I think in no small part that's also because their customer support is absolutely excellent. Press F1 twice and you're in a chat with the helpdesk; these days the initial query is triaged by AI of course, but for actual issues you're talking to a real human in less than a minute (sometimes just seconds), and these real actual humans typically have the expertise to help you. If not, the higher levels of support are good too. Don't underestimate this aspect!
Yeah, my reply’s a bit tongue-in-cheek. I don’t really oppose monopolies per se – it’s a free market, a better product could compete and win.
But one of the Bloomberg’s main selling points, at least from the last discussion I’ve read on HN (and some of the comments here too), is the chat, which means there’s network effects at play. So you can’t just build a better terminal, you also have to convince the majority to switch.
(Though apparently Bloomberg Chat also supports email? Maybe the competitor can use that to interface with it, so it’s only “green bubble” bad, not “all your colleagues are there and you’re not” bad.)
Ah, I see and didn't pick up on the tongue-in-cheekness!
The network effect is real and I don't think you can interface external email with it. You do get a Bloomberg email address as a user but I've never seen anyone in a chat with an external email only.
Maybe it's because of all the integration that's hard to do. A bit like if Windows 98 was inseparable from Internet Explorer, so you couldn't just replace one or the other.
For instance, Bloomberg provide you with:
- real-time data on financial instrument prices;
- real-time news (own and syndicated);
- an email address (firstname.lastname@bloomberg.net) used to email and chat with other professionals on the Bloomberg network. This is like LinkedIn and WhatsApp combined for people in finance, and people can keep the address for two years after e.g. losing a job, so they can use the network to find a new role.
- analytical software: tools to calculate various KPIs to assess and analyze business opportunities.
- capability to transact: The BB terminal is not just an information system, it also permits you to buy and sell directly.
(My university has one BB terminal that MBA students can access and get training on.)
It's all that and much more. It has an enormous amount of data of all kinds that could be vaguely relevant in finance. BMAP can show you vessels (AIS data), location of LNG terminals, powerplants, bank branches, crop yield forecasts (just a few I picked off the list just now) and so much more on an interactive world map.
And every kind of price you can think of (almost). Crack spreads? River levels? Baseload power? Airline schedules? I am trying to think of more but it's truly VAST.
They have custom screens, configurable screens, custom settings (you want the generic futures contract to be rolled differently from the default? Sure...), option pricers and stuff I have never had to search for or look at.
They also have their own query language BQL for data queries and the BQNT platform is Python-based and optimized for use with the data.
If you just want the latest stock prices for your portfolio your online brokerage app probably does a better job.
The biggest miss right now is their ASKB AI chatbot, which is borderline moronic. It has to be some older model with low thinking/effort, because the first answer it gives is always broken in some way. Hallucinated functions or fields, outright wrong descriptions, etc. I don't even try to use the first answer anymore but ask it again to check its previous answer against verified and authoritative sources/documentation and will come up with many corrections. And then I do it again, and maybe a fourth time.
> I don't even try to use the first answer anymore but ask it again to check its previous answer against verified and authoritative sources/documentation and will come up with many corrections. And then I do it again, and maybe a fourth time.
Maybe all they need to show massive improvement is a for loop of four iterations with the loop body being "are you sure, check the docs?" and return the result at the end.
From what I have seen of it - for the people who can afford it I suspect that having everything all in one place under a common UI is a bit of an enabler.
Reducing the friction or feedback loop of information retrieval/analysis can change how people work just like repls and hotreloading did vs slow and long compiles
I've heard many people quip that the best thing about using a $25000 terminal is being connected to other people who can afford spending $25000 on a terminal.
Its breadth and depth - it’s about being able to see a forward for ISONE Vermont Pk power prices, swaptions on uk gilts, where 5yr forwards are pricing for the yen, freight rates and ship tracking, weather data, satellite analysis, legal and legislative data etc.
You hit HELP HELP (a dedicated button on the keyboard) and a real, knowledgeable person in your time zone will _fix_ any problem with the data you see on your screen. If it's not fixed by the time their day ends, they will hand your ticket to yet another knowledgable person to the west who will continue working on it to fix it. That is, the data quality, and the data quality processes, are a large part of what you pay for. People get hung up about real time, but what if a number on a financial statement from 10 years ago is screwing up your model training?
Some markets, for example, corporate bonds, are mostly on BB chat. They are dealer markets. And dealers will provide “runs” of prices on chat. List of 200 bonds and their prices.
So if you want to trade bonds, need BB as that is where it is happening. Keep in mind some of these bonds trade infrequently (sometimes few times a month).
Having said that, there are more and more bonds trading electronically all the time.
They have some public demos! https://youtu.be/ZAYiEyaUk2M?t=143&is=fXjn7PcE47Dw2Ayf
I’m pretty sure you can get a one month free demo (with limited data feeds).
Yeah, I think this is something a lot folks miss when doing UI/UX these days. There’s an immense focus on new users, not on folks who rely on it day in and day out for actual work. I’ve seen loads of old, incredibly “ugly” UIs that absolutely mog newer friendlier attempts at the same. They’re absolutely intimidating to new users, but once folks learn it they’re indispensable and obviously intentional.
I really wish there was a resurgence of information dense UIs and tools, optimizing for people who use them daily and not “use it once.” It really reminds me of the hate folks have for things like Eclipse or IntelliJ which are daunting at first but once you get over the shock and learn them really are amazing tools.
You’re reminding me of this thread I found a long time ago where someone was talking about the old amber-screen library computers that felt FAST. The gif here shows what it would look like to read Twitter in that manner… https://x.com/apaprocki/status/1230241842928791552
Well there's a name I haven't seen since 2007
Hope you’re doing well! A lot has changed since then :)
https://x.com/scottew/status/1751357591375208689
The difference is that Costco pays their employees well enough to keep them around (notice that Costco prints the start year on each employee badge- average retail length of service is 1 year, Costco keeps people around for decades) and so they can learn to use this older software and master it. Most retail operations optimize their inventory management for new people, because they don't expect someone to stick around long enough to master it.
My wife worked at a small regional chain of pharmacies that had an ancient MS-DOS based EHR system. And she mastered that system to the point where people would call her when she was home with me and ask her how to do something, and she would walk them through "okay, hit tab three times then shift-F12 then enter the frabble and then alt-F3 to go to the next screen and enter the drabble." It was difficult to learn it, but if you mastered it you could do everything you needed to do from just muscle memory while you were on the phone arguing with the doctor's office that their prescription was not correct and needed to be changed.
Then the small chain got bought by a national player and their brand new GUI system was brought in. It was a lot easier to learn- discoverability was a lot higher with menus and drop-downs and all of that. But if you stuck around to master it, you could never be as fast as you could with the older system. And that emphasis on quickly integrating new employees- but lower payoff for experience in employees- was clear in every choice that national chain made. They wanted their employees to be completely interchangeable, which meant that they didn't care if you had six months of experience or four years of experience or three decades of experience.
Same story at my bank. Tellers used to speed run through a TUI, hitting tab and other key combos to accomplish things. Now it's a long wait with them mousing around.
The best thing about all character-based applications I've ever used is the keyboard buffer. Even if you're using a TUI on a system that has a slow backend, it almost always blocked the thread and didn't consume from the buffer during waits, so if you knew the sequence of keys was "Total, F1, 1, Enter, Enter, Tab, 2125551234, F10, 1, Enter" you could just enter all those keystrokes as fast as you could and the system would eventually catch up with you, meanwhile you were bagging a purchase or counting out the cash, whatever.
The GUI also has lag - especially in the era of web-technologies-bloat, more lag than ever - but whether touchscreen or mouse there can't be a buffer, so each lag requires you to stare waiting for the next UI to be drawn.
Unfortunately web-inspired TUI toolkits start to lose that property. Hit a key too fast, before the next screen runs code to attach its event listener? Sorry your key is lost.
Clickable web GUIs that don't lose your interactions are even worse. The email client I have to use at work buffers your "mark as read" click when it's loading new letters and applies it to the one that's under your cursor when it finishes loading them.
This is a good point: you can build character-oriented interfaces "for dummies" — IBM i née OS/400 which Costco uses is actually good at this, with a standard interaction design centered around menus and form interfaces that map directly to CLI commands — and information-dense GUIs, but you can't very well build point-and-click interface with keyboard-style buffering.
Agreed. Also, you can try to satisfy both new and expert users. It take more thought and design, but that principle is in both Nielsen's Heuristics and Shneiderman's Golden Rules.
> Most retail operations optimize their inventory management for new people, because they don't expect someone to stick around long enough to master it.
Somehow I think I failed to think very deeply about this very important link! The twin happenings of the elimination of any retail career that pays a living wage, and the availability of GUIs and especially touch-screen GUIs happened simultaneously over a couple of decades, and the touch-screen dummy interface naturally won out. Not just because training that used to take two days for the 'green-screen' system I learned in 2001 now takes 15 minutes of instruction (and even that training is rather redundant for anyone over 95 IQ) -- also because the massive differential in productivity of an expert vs. new cashier/teller/pharmacy tech/etc. only even manifests if people are on the job long enough to become expert. And they're really not anymore.
Kinda why I prefer self-checkout. Even though the UI is firmly in "dummy" territory and I have to use a touch screen, I've used the current iteration of the self-checkout software more transactions than most cashiers I meet at Target, since they seem to have been there about 2 weeks on average.
The de-skilling of retail clerk I think predates even the GUI. I think it goes back to the laser/upc/database systems of the 1970's and 1980's.
In 1970 retail clerk is a medium-skill medium prestige job. There were lots of stories of people who started as a retail clerk and end up as CEO, and even people who aren't that ambitious could still carve out a comfortable middle class lifestyle as a retail clerk. But that's because there was a reward for experience. Before the laser/UPC/DB system rolled out, you needed at a glance to be able to know the price of an orange, or the difference between the expensive and the cheap mushroom. Because there would be a paper tag on every cheap mushroom that said "10 cents", but what if someone swapped that paper tag onto the expensive mushrooms that were supposed to cost 30 cents? You needed to know the difference, and have the real prices memorized to be able to be fast enough to keep the customers happy.
This also meant that you couldn't have a large retail store (something at the scale of the modern Walmart or Target), because you couldn't get enough clerks who could remember all the necessary prices and so inevitably you would lose money to tag switching. Department stores solved this by forcing you to buy each good separately in each department- you would pay for your housewares in the housewares section and your clothes in the clothes section, and they had another whole army of clerks in the basement managing store credit to keep it low-friction enough. But that's very high-touch, demanding a lot of workers, and those people still have the rewards of experience so they are all comfortable careers.
But then the laser/upc/database system came out. And suddenly you would get a natural language description of the item (or even a picture in the modern systems) from scanning the UPC. And you could scale the store up as much as you wanted, just needed to buy a bigger hard disk for the database, so the stores tilted more towards capital and less towards labor. And that eviscerated the career path of most retail stores, because store owners no longer believed that keeping retail clerks around benefited them.
This is machines hallowing out a job and alienating workers from their labor for the benefit of capital- exactly what Marx had described a century earlier for factory workers in Das Kapital. But for reasons of class and gender we mostly ignore retail clerks so it doesn't seem to have attracted near as much attention.
I think that's the 'push/pull' problem though. Information dense UI that takes months or years to learn, or something that may be less information dense, but a monkey can learn it in minutes.
For me, it's typically a balance between speed and efficency of data to help you make decisions quickly, but should be able to be used by anyone with technical experience.
So less of both. Less 'pretty UI' and less "omfg here is every piece of information shoved together and you need to be a wizard to understand it"
The same for hardware. Where I used to live, the ticket vending machines had you punch in your 3-digit station code, adult/child, single/return, and boom ticket issued. You had like 6 button pushes and could buy a ticket in seconds.
In this video I made the mistake of using a debit card (still very slow to clear in 2005) which adds the delay https://youtu.be/pQCktMWR94Q?t=7
They replaced them with Windows-based touchscreen vending machines where each screen selection took as long as the whole former selection process took just to load.
I love displays from space vehicles and nuclear plants for the same reason. Data and controls organized in a dense but presumably highly usable manner. And it covers the more specific view of an expert in a particular subsystem, as well as the higher level view of those responsible for the whole system.
I love NASA design of their space probe telemetry/control views.. lol.
It's so pure.
I remember seeing glances of the Mission Control screens during the Apollo and early shuttle program - two screens and two channel selectors where the programming would be a screenful of real-time data rendered in the adjacent RTCC (real time compute center, or something like that) and captured by slow-scan CCTV cameras and piped to the consoles as TV signals.
Also on the input side: for instance here is a cheat sheet for a small part of the available keyboard commands: https://www.bbk.ac.uk/documents/bloomberg-cheatsheet.pdf
Yes, but where are you going to place ads?
Nobody gets a promo from something that is working well for humans.
I wish we got more UIs like this.
Most modern apps are very lazy on being actually useful for people that use it frequently
UI designers who were trained in the currently-popular paradigm would be at a complete loss to design such things. "Swipe left from the home screen, tap the nothingness on the screen to reveal the controls overlay, find the ••• in one corner of the screen, tap "More" in the 3-item menu, and then tap 'Bond Quote' in the modal dialog that takes over the entire screen." Keyboard shortcuts? "Hmm... interesting idea. Never thought of that. We might add those in a V2."
My god have you ever had those people try to add keyboard shortcuts? You start to question if they have ever used a keyboard at all.
We already see this in car interfaces. Clearly designed by people that don't drive, or at the very least have not tested it themselves while driving.
I got a rental the other day and while driving could not find the clock! I had to hit TWO BUTTONS on the infotainment touchscreen panel to get back to "home" and see the gd time. I was driving on curvy roads in Yosemite too. Truly abhorrent.
Also bad: When a software update moves essential elements around.
For the last three years I've looked at a particular location in my car for the clock. Then a software update moved it not just to a different location on the screen, but to an entirely different screen.
Three years of muscle memory is probably going to take another three years to un-do.
(For those of you who don't understand why a clock is an essential part of driving, it's so I know if I can park in a 9a-4p zone, or if it's the time of day when I can use the reversible lanes in my direction of travel, or if I can use the lanes that are HOV sometimes and HOT at other times, or a hundred other things.)
The same software update moved the wiper locking controls (for car washes) from a button on the left steering column stalk to a menu option four levels deep on the center screen.
Oh wow. This would make me lose it. I've been increasingly angry lately with enshittification. It's so frustrating and unfair.
I got a rental car recently and discovered there is no way to change the speedometer into kph
That's definitely surprising in this age. Could you change the air conditioner ui to celsius? But changing to kph is not something I expect by default in _every_ car. I do expect a visible clock in every car though.
Actually yes I did notice that was an option, changing the units for the temperature was there.
> have not tested it themselves while driving.
I don't think the driver is supposed to use the touch screen while the vehicle is in motion
The corollary of which is that no function the driver might sensibly need while underway should be presented or accessed through a touch screen.
So physical levers and buttons. With some tactile feedback, such as raised lips or centre bumps or distinct shapes, so without looking the user knows which one their finger is on (third from left in the Aircon group). Displays can be present multiple items (don't need physical dials) but changing that display when driving comes back to buttons.
The grenadier interior is supposed to be great in this respect. Sort of looks like a cockpit
https://www.caranddriver.com/photos/g63743843/2023-ineos-gre...
Apart from the aesthetic style, that's just what cars used to look like. A bunch of buttons for various things in the middle.
Which is the problem, yes. It was perfectly fine to use physical buttons while driving because you don't have to look at them.
The law in the UK appears to be that is is perfectly OK to use a touch screen so long as you "stay in full control" and aren't holding the device:
https://www.gov.uk/using-mobile-phones-when-driving-the-law
NB My car has loads of features that I would expect to use while driving that can only be accessed through the touch screen.
If that's correct, why is it there at all?
(increasingly everyone including the safety wonks are realizing why this is a bad idea)
My car has user interface elements that you would want to use while driving (e.g. to cancel the navigation guidance) that only become visible if you hold your hand in front of the screen....
My car display has NO WAY to cancel navigation guidance from the GPS screen. You can pause guidance, but when you turn the car on again, it tries to resume.
You have to go back from the guidance screen to cancel. So, I never use it.
Because Bloomberg Terminals are specialist devices for power users and most apps/web sites are consumer products.
Don’t get me wrong, I get the complaint. But it’s not like the designers are unaware of more information dense alternatives.
> But it’s not like the designers are unaware of more information dense alternatives.
Strong disagree. I have been in UX reviews for products like Bloomberg terminals as the intended user. I specifically asked for information density and the designer looked at me like I was from the moon. Their counter offer was a permanent floating chatbot that covered the small amount of information still on the page. No joke. Information density is something those web designers genuinely could not comprehend.
“ You can remove the drawn-in borders just by adding a little visual space, and use subtle color differences to separate rows if needed” ———> “Let’s make it 10 rows per screen (and somehow still with low-legibility font choices and low-readability text design)! Make the design breathe, you know, white space!” ———> Me, having an aneurysm
I described it as being lost in an ocean of white pixels. Then they added padding and broke hyperlinks. I know it was incompetence but it started to feel malicious.
Saw this post today about ceiling fans.
what is going on with ceiling fans: https://mcmansionhell.com/post/829127919552151552/what-is-go...
https://news.ycombinator.com/item?id=49917536
I would love to read up on how such interfaces (spaceship flight controls, avionics controls, etc.) are designed and the history of them. If anyone knows of any such resource, please point the way!
This is the difference between accessible and efficient. Bloomberg, and aircraft controls are expert interfaces, they have steep learning curves and in the latter require case extensive education to use (Bloomberg also does a ton of work on user training). Which is absolutely fine when efficiency trumps everything else and it's a system people use all day every day.
Most interfaces we use on a day to day basis are the opposite, they're designed to be accessible (in the sense that you can use it if you are unfamiliar, not disability) for someone unfamiliar, untrained, and uninterested in learning, which also aligns a lot of the time with sales growth. Nobody is willing to do a 6 week course to use the new HR UI really, really efficiently. The unfortunate thing is some of these are enterprise UIs some of us do have to use all day every day and that's where the impedance mismatch gets painful.
> Most interfaces we use on a day to day basis are the opposite, they're designed to be accessible
I read this and then I think of glass doors with handles on both sides that I don't open till I read the instruction on the affixed plate: PUSH/PULL. The door forces this decision on every single conscientious user. These things are everywhere, including software interfaces.
As a point of interest, the PULL handles are supposed to be vertical, and the PUSH handles horizontal.
But that doesn't always look nice on a glass door.
Was going to say almost exactly this. Very similar experience in my background as well (CAD interfaces).
Accessibility peaked around Windows 95 and Windows 98 when buttons were clearly buttons, scrollbars were clearly scrollbars, and clickable text was in blue and underlined. We're now well past that, and without familiarity with disappearing scrollbars, three-dot menus and other obscure means of navigation, it's very unintuitive to get around. We don't even have manuals anymore if you do want to learn how things are supposed to be used.
Strongly agree with you. Doug Engelbart (of mouse + groupware + hypertext fame) once said that if ease of use were the only design criteria, we would all ride tricycles instead of bicycles.
Not every user interface has to be or should be walk up and use. But there also needs to be a clear payoff to training.
> Most interfaces we use on a day to day basis are the opposite, they're designed to be accessible (in the sense that you can use it if you are unfamiliar, not disability) for someone unfamiliar, untrained, and uninterested in learning, which also aligns a lot of the time with sales growth.
Must be why my fghxjing iOS jorard doesn’t work, and why searching for “preferences” (which is what it is called on Mac) doesn’t open “Settings”. Bonus points for gazillion of gestures that require the Necronomicon to decipher.
I wonder how "make it Bloomberg terminal" would change the ui outcome from ai...
The modern Terminal is based on a private fork of Chromium to give it the look and feel of a VT100 terminal, and integrate their private networking and security technologies. The BBT predates HTTP, and backwards compatibility is hugely important to the company- they have a museum where a second-generation Terminal from ~1985 shows the current news, because they are so dedicated to backwards compatibility that they still can support the 1985 hardware.
This isn’t really true. The old terminal monitors are essentially dumb CRTs and the the museum pieces are just displaying the output from a regular PC running the modern app converted to an appropriate signal + BNC connectors so you see something and not just a black screen.
The old terminal hardware communicated over dedicated serial lines and worked in conjunction with many other “server-side” pieces of software that no longer exist, so it would be impossible to make one function as it did back in the day.
I would love to learn the specs of the original 1983 terminal. By the look of it, it's NTSC timings, so 200-ish lines, but it shows text and graphics, so it must have some bitmap capability. If I were to replicate the look, I'd use something that can display ReGIS or NAPLPS.
They ran iRMX on an 8080 variant and had graphics card that handled the A/V aspects. I suspected the displays could have actually been PAL because it had slightly higher resolution, but I haven’t been able to confirm that yet. I have one of those “Smithsonian” pieces on my desk waiting for me to have to some free time to try to repair the CRTs when I someday have more free time than I do now (any good resources for that??).
In these days I’d as my friend Claude.
Hey Andrew, former BB employee here. What's the dev experience like over there now? Back in 2011 when I left, RnD was leaning heavily on the custom JavaScript + IDE for frontend dev, and mostly cpp and non-relational mmap DBs on the backend. How have things evolved?
So much has changed since that time. There's a great recent talk by Jason Williams and Paul Williams that goes into a lot of detail [1]. Most things now use TypeScript developed natively in VS Code using extensions and a CLI for the additional proprietary bits. That all nicely integrates with agentic workflows, skills, etc. ComDb2 is very relational, very battle-as-in-jepsen-tested and has been open-source for a long time now [2]. C++ is still king, but there's plenty of Python, Go, Rust, Haskell, OCaml, etc. running around.
[1]: https://www.youtube.com/watch?v=y1MCLZm8yAY
[2]: https://github.com/bloomberg/comdb2
Ok, lemme ask you this: What has been your favorite Bloomberg keyboard over the years? I preferred the mid-2000s mechanical keyboard before they started going lower profile keys, which suffered from off-center binding when pressing near the edge of a key.
> VT100
Haven't seen that term in so many years/decades. Thanks for the flashback.
I got my eyes checked a few years ago and they were literally using a VT100 interface. The person knew the keyboard interface like the back of his hand and it was so much faster than a mouse-based GUI that is so prevalent.
Every terminal window these days is an emulated VT100.
No, not really. Most likely something between a VT-220 and a VT-525.
For completeness' sake, here's also a history of competitor REUTERS' terminal:
• https://www.thebaron.info/archives/technology/reuter-monitor...
• https://www.thebaron.info/archives/technology/reuters-techni...
Bloomberg's interface looks overwhelming at first, but a trader who knows where everything is can scan a lot of information very quickly. There's probably a useful distinction between an interface being diffucult to learn and an interface being inefficient to use once you have learned it.
Starting October 14, 2026, Bloomberg is requiring a physical Bloomberg Keyboard for login and access on all Bloomberg Open Terminal workstations.
The harder Bloomberg tries to control and restrict its users, the more users will just find alternatives or even build their own proprietary systems. It thinks it can be Apple in the financial data world, but I highly doubt it.
Wrong take. Bloomberg has a data monopoly on certain asset classes, most importantly bonds.
I doubt that's a big blocker - the keyboard is one of the selling points and I'm guessing that you could just have 2 keyboards if you really wanted to (the Bloomberg one for auth and the other one for longform typing??)
The current Bloomberg terminal keyboard actually feels nice. It looks cheap, but feels decent.
And has all the right buttons.
Maybe related, but is anyone familiar with a series of trading terminals from BT in the UK? I cut my teeth in circuitry repair working on these models like the 405 and 405i in the early 2000s. These days I can't find hide nor hair that these things ever existed! Any time there's an article that mentions the Bloomberg terminal I always have dig around for this mythical series of trading platforms, alas I now worry my memory is playing tricks on me.
I wish that article showed what the display looks like instead of just the keyboard. Others here have mentioned the lovely information density; I'll have to search elsewhere to see it
Related: A look back: The Bloomberg Keyboard
https://news.ycombinator.com/item?id=17079033 (104 comments)
Thanks! Macroexpanded:
A look back: The Bloomberg Keyboard (2017) - https://news.ycombinator.com/item?id=20742329 - Aug 2019 (55 comments)
A look back: The Bloomberg Keyboard (2017) - https://news.ycombinator.com/item?id=17079033 - May 2018 (104 comments)
> joined by former Salomon colleagues Thomas Secunda, Duncan MacMillan, and Charles Zegar
This is just too funny to be true: one bloke named "a second" and the other one named "clock" in Polish language
Andrew Paprocki: Bloomberg: home-grown server-side scripting
this is a great video to watch if you're curious about how the bloomberg terminal ui was made. https://youtu.be/ODgs0eWAIKc
…also, frequent commenter in this post!
It may be of interest to some that Bloomberg has an ocaml based DSL built in for pricing exotic derivatives.
Ocaml at Bloomberg, https://news.ycombinator.com/item?id=28278553
175 points on Aug 23, 2021 | 138 comments
OCaml is visible in a lot of technology used daily by many engineers at Bloomberg.
Right but it's also exposed (I'm not sure if it's actually pure ocaml but it is presumably a form of it) to users
Where can I find it in the terminal? Sounds like a fun thing to play with...
DLIB BLAN will take you to the playground
Thank you, very cool!
Does someone know any font that is similar to the closed-source Bloomberg terminal font? I really love that font.
Some backstory on the font that Matthew Carter designed is here: https://www.bloomberg.com/company/stories/how-bloomberg-term...
Some people like it and projects like this have appeared: https://github.com/dchest/bergoom
Thanks! bergoom looks promising - no monospaced variant, but I think that can be done.
I was working there when they switched to that font. A lot of people were furious that their cheese had been moved.
I did an internship in finance in college. Worked at a Bloomberg terminal that earned more money than me
The article has an image of terminal keyboard used by a legendary trader... the key-caps are immaculate; the letters haven't faded at all. Amazing. I wonder if they were double-shot...
On one hand, at the price point of >$10000 for the whole terminal, it's quite possible that >$100 went into the keyboard.
However, I suspect you're coming at this from the perspective of a coder or a gamer or some other category who types thousands of keystrokes per day (tens of thousands?). With the assortment of special-function keys that enabled the user to execute common tasks with only two or three keystrokes, I suspect that even actively-used Bloomberg terminals may see an order of magnitude less keystrokes.
(On the third hand... I seem to remember reading that the Bloomberg system included chatroom functionality, which may increase keystroke count)
Especially if you're the BOSS - you may have a terminal, but you may transact by telling your underlings what to do on your behalf.
> I especially love that Gross taped his login and password directly on his keyboard, which makes the object more relatable. I may never know what it’s like to manage billions in assets from a Bloomberg Terminal, but I absolutely understand the trial of remembering my passwords.
Wow, what a great history artifact. It is hard to look at this and not wonder where did we go wrong. Everybody _should_ know better, and yet here we are. And now immortalized in the National Museum of American History for all posterity.
Any idea on what technologies/toolkits are used to build Bloomberg terminal?
Nowadays the desktop software is built around Chromium. Lots of C++ and one of the largest JS codebases in the world.
> largest JS codebases in the world.
I'd love to know how they make frontends as responsive as possible. To the engineering blog!
Here’s a great talk to get you up to speed! https://youtu.be/y1MCLZm8yAY
Super interesting, thanks!
It's not very responsive
From memory when I worked there 10 years ago, all the Javascript based stuff was pretty slow in a very noticeable way. They've likely gotten better as they are using a well known Javascript and rendering engine, so they're getting many optimizations for free.
They're very involved in web standards. Rob Palmer from Bloomberg is co-chair of TC39.
We’re in the middle of TC39 plenary in Tokyo right now as a matter of fact! Rob regularly posts on X as things advance through the staging process, and today is the final day.
Agenda: https://github.com/tc39/agendas/blob/main/2026/09.md
It used to be built with qt, now I'm guessing the answer is vanilla JS and CSS. You'd be surprised how fast they are without frameworks like react
It has never used qt in any way.
Love how those dense screens make numbers checkable at a glance. Tiny shops still live that gap—bank lumps vs invoice lines—just without a dedicated terminal.
A keyboard that resembles the old-school bloomberg one would be so cool. Wonder if something like that already exist...
The current models are just USB keyboards with special buttons. The requirement of an official keyboard will increase the second-hand market prices, but they used to be available on eBay.
Have we made any progress on a cracked version?
https://www.reddit.com/r/wallstreetbets/comments/8aiz67/how_...
it's saas with authenticated market data.. there's nothing to crack
The best line I found there:
> They turn your money into information or something
Zitron must get on this ASAP it will make his day
Thank you
Godel > Bloomberg
Otherwise known as "that thing beside Danny DeVitos' bed in the movie OtherPeoplesMoney... i want one of those!"
At $32,000 a year per person the Bloomberg Terminal is a deeply unfair advantage for the rich to get richer.
The entire financial industry is based around unfair advantage.
Capitalism doesn't care much about "fair". It cares exclusively about efficiency.
Yeah, but I think they just sell convenience and there are projects on GitHub like openbb that piece together something very similar for much less.
In finance, people who consider the cost of a Bloomberg terminal "expensive" are often considered "not serious" (small fish)... it has often been seen as a status symbol as well as a professional work tool.
Not really. If you ran a restaurant instead, you may also have to buy a few thousand dollars worth of ingredients and pay thousands of dollars in salaries per year and renting a venue to make the revenue that you are making. Even a pizza delivery service may need a few thousand dollars a year for a car and its driver plus petrol, taxes, insurance. Relevant is not the cost of doing business, but the annual revenue and margins.
E.g. $32k cost of business to make $2bn revenue statistically approximates zero.
I googled the ovens used by my local pizza joint and they were 25K each and there were two of them.
Heck, the espresso machine at your corner Starbucks probably costs almost that much.
that's a perverse way of looking at the cost of a service offering.
It's the very nature of our society that money buys you increased access to tools for making more money.
Well, what would be the point of being rich, if you couldn't buy more stuff?
I swear, the audience around here these days...
> Well, what would be the point of being rich, if you couldn't buy more stuff?
One could ask "Well, what would be the point of having power, if you couldn't abuse it?"
One could, but it would be a completely-different question.
Having a leverage isn’t an abuse. It’s a tool, it’s not like you’re buying a news outlet or a local senator.
If that were true, it would cost more than a Toyota RAV4
youre assumming that the average person can use bloomberg profitably. thats absolutely not true. the average person gambles on crypto or polymarket
Not to sound like a YC fanboy here, but I love how pretty much every great business can be traced back to a start which somehow follows what YC talks about today.
I guess the first principles of what makes great Founders and great businesses from back in the gilded age to today have stayed the same.
Can you expand a bit more on this? What first principles?
shit I really loved this, was searching for the history of bloomberg these days. thanks for sharing