Hacker News new | ask | show | jobs
by lilott8 2402 days ago
Excel is the most widely used domain-specific language -- maybe even widely used "programming language" (if I'm painting with a broad brush, though it is turing complete) used to date. Getting people off, when a company's livelihood is literally at stake, is no small ask. Solve this problem and you'd be a bajillionare (official title).

I scan the comments and see a few anecdotal comments about how life is just fine with LO. But as many other comments point out: it's not the one-offs that make this hard, it's the fact that excel is the de-facto information exchange between people, businesses, and boards. To be a broken record: it's too ingrained.

11 comments

The CS industry doesn't understand that the rest of the non-CS world runs on tools the employees learn once and never look back.

Coming from engineering, CS people never seemed to understand why someone used Matlab at all, when python existed or even the utility of 'R'/'Stata'.

The effort needed to onboard onto a programming/mathematical/computational tool, when you don't have a strong coding background to go with it, is extremely high. The users will hold on to them till their dying breath, because the is tied to more than 50% of their own value proposition.

Excel is the epitome of this phenomenon.

Also, excel is straight up a good tool. The only advantage of Libre office is the price, which is a non-factor for any major corp. Google's collaboration suite is better, but they lag behind in every other area.

I can see windows being completely replaced by competitor, but Office will stay. It is MSFT's stickiest product.

> The only advantage of Libre office is the price, which is a non-factor for any major corp.

Another one: Excels stubborn insistence on mangling anything that can be somehow construed into a date into one :-/

Format your column prior to pasting data, paste special > match destination formatting
the most solid way to handle it is to use data import from text file (handles csv) so you can configure the columns while looking at them

Which recent excel versions you can bring that data table directly into power query as well, which is useful if youve got multiple csv you are trying to look at that are related

> onboard onto

Another way to say this is: "learn".

i think 'onboard' might refer to the wider context of using a tool in say, a research group. it's not just you learning the tool, it's about how the group uses the tool and the historical developments in place.
It's just a tech business language mannerism. People use it for the same reasons they use "price point" instead of "price", or "usecase" instead of "use", or "form factor" instead of "size": to show that they belong to the tribe, and to sound clever while saying something that could be expressed more simply and elegantly. The jargon adds nothing, except cringe.
Another way to say this is: "con". But language changed. It's almost as if it's defined by what's spoken, not by what's in a dictionary.
Another way to say this is: "teach".

People who code for a living are used to regularly teaching themselves new things, but there are lots of folks in other corners of the organization that expect to be taught / onboarded, and it's not an entirely unreasonable expectation. I think the word "onboard" implies a cost will have to be paid (e.g. time, money, man-hours, hiring new talent, etc.), and that's what the parent poster wanted to emphasize.

worked with a lot of techs this way. They get mad if you show them what you're trying to teach them shows up with very good instructions on the first page of google. At least half of them get pissed off :)
You’re right about Office being sticky, but that is changing. Heaps of people are using Google Sheets for basic spreadsheets these days. Advanced users will stick to the tools they know, but I’ve reached the same conclusion as the author of this post. Licensing issues, nagging about cloud subscriptions, inconsistencies between Windows and macOS and the lack of Linux support have become so painful in recent versions of Office that using LibreOffice no longer feels like a greater inconvenience – just a different inconvenience.
At least LibreOffice runs on Linux/Windows
In my experience anyone in CS who was doing enough work for which Matlab and R to be relevant definitely found them useful, at least in the correct situations.
Proprietary math software, to which I'd add Mathematica, has the specific non-value proposition that you cannot take home your precious research work from the university lab without expensive, hard to buy and impermanent licenses and subscriptions.
Take all of this with a grain of salt: I'm an Excel and VBA developer. I'm a bit of a dinosaur and I'm looking to pivot to Java + JavaScript as I'm just a smidge underpaid.

However, I have this to say about Excel. MicroSoft is making a push to kill VBA with two prong. One, replace VBA with JavaScript through OfficeJS. Two, beef up Power Query and Power Pivot so that you can do enough ETL in Excel through a point and click interface.

But the nasty truth is that on the ground in offices worldwide, there are ETL flows that shouldn't exist, but middle managers do not possess the political capital, incentives, or technical skill to remove. And while Power Query, M, DAX and Power Pivot are excellent, in, let's say, 5% of workflows, you need some business logic in some language that is more flexible. And this is the real problem with VBA; it papers over dirty workflows, even if the data structures you are provided are not good.

What kind of software do you develop in VBA? I've used it a fair bit over the past few years to build spreadsheets and Access databases for people in locked-down enterprise environments who can't run arbitrary .exes. I wish I had just built web apps, despite the huge downsides of needing to trust a system administrator to secure user data, and being sandboxed into a browser.

Using VBA was much worse. I could manage with the VBA language alone, but the number of security controls and macro-related warnings I had to click through, the poor quality of free online documentation, the lack of version control support and the hacks I had to use to get things into Git, and the lack of decent libraries made the experience an overall nightmare. Have you found solutions to these problems? Or do you work in an environment where you control the users' computers?

There are other areas like this, too. I'm part of a firm that does grant writing for nonprofits, public agencies, and some research-based businesses (http://www.seliger.com for the curious). Word and Excel are standard, especially for any documents that may touch governments. All narratives are written in Word and budgets in Excel.

I like and approve of LibreOffice but it doesn't seamlessly transfer complex styles or tracked changes to and from Word. When an organization is working with people who bill in the hundreds of dollars per hour and a grant in the hundreds of thousands or millions of dollars, cost of MS Office is a non-issue. I approve of and generally support OSS (I'm writing this in Firefox), but the level of integration into the MSO world is ill-understood.

I've also written novels and the whole publishing industry and infrastructure is based around Word.

This is very true. Fighting Excel is like waging the Drug War. You can make what appear to be gains while still losing.

As someone who started in Windows 97 and .NET at the 1.1 Framework release: VBA is a gateway drug to a career in MS languages.

There has never been Windows 97.
You know that was a typo.
I don't think a lot of businesses are really that dependent on the sophisticated Excel features. And if they are it's concentrated in certain areas. You can buy per seat licenses (honestly no idea how MS prices this) for whoever needs it and out the rest of the enterprise on something else for their documents.
But why should they do this? MS Office 365 is 150 USD per year per user. For the vast majority of companies, large and small, this is a very small price for a software that is so engrained in daily business processes and activities. If a company is at the point of deciding about cost cutting by cutting excel, it probably is beyond salvation (certainly in the developed world, but i dare say even in developing countries). And the productivity loss of the first year would probably negate any saving.
> But why should they do this? MS Office 365 is 150 USD per year per user.

As someone in the Linux/Debian/Ubuntu area of my organization, the main headache with licensing from what I can see as an 'outsider' is not so much the price as the auditing/compliance overhead.

We can spin up/down as many Deb/Ub/CentOS systems as we want and not care, which as much RAM and (v)CPU as needed.

Trying to do the same with Windows and/or VMware seems to be something else.

Absolutely. 150 USD is a trivial amount of money in most bureaucracies, but the time cost of getting that expense approved and accounted for could be weeks.
Excel is sold as a bundle with Office 365. Many if not most large companies are already paying for O365 for emails/teams/MDM etc anyway.
My company is standardized on G Suite and only dole out MS licenses on request.
Thanks for the anecdote! There does seems to be diversity among the suite of tools large companies use to conduct business.

The parents point seems to still stand -- a significant proportion of companies use Office internally, and Excel is bundled as a part of O365.

To add another anecdote, I've noticed that many educational institutions (including government departments with tens of thousands of staff and hundreds of thousands of students) use G Suite. I'm sure some corporate staff in these organisations have Microsoft Office too, but there are many people in large bureaucracies who use Google Docs and Google Sheets daily.
The excel work I've seen finance gurus perform is absolutely programming. There's usually "that guy" who assembled it and understands how the massive 100 tab completely uncommented spaghetti mess works.
And then he retires or gets run over by a bus and I have to spend three months reverse engineering it only to find that it is really just an over complex UI hiding a bug ridden home made database management system. Which I then reimplement in SQL Server and C#/VB.Net in a week; and the new version is faster, properly multi-user, more reliable, and actually maintainable.
Nah thats when you hire a consultant to support the spreadsheet at twice the rate. You can even pay them to make a new spreadsheet and train the next unsuspecting victim.
That's in fact what I was hired to do. I told them the first day on the job that they really needed a database, even sketched the schema. But they were adamant that it had to be an Excel solution so I wasted three months trying to untangle the crap that was there all the while trying to persuade them that a DB and a simpler front end would be much better. Eventually they agreed.
We've both seen too much... O_O
..and people wonder why Google hasn't pulled the plug on Chrome OS and Chromebooks. It is essential for the next generation to prefer "Sheets" over "Excel".. only way to do that is to get in early.
That hasn't appeared to work. Students who are forced to use Chromebooks don't beg their parents to get them one for home. The ones who do care about laptops ask for PC's or to a much lesser extent Macs.

That also never really worked when Apple was much bigger in pre-college education.

Edit: Chromebooks in education are like Enterprise software. The end user never chooses to use it. The vendor is not incentivized to make a product that end user will like, only the corporate buyer.

How many teachers and administrators who do have a choice are using Chromebooks willingly?

Many. I’m on a K-8 school board and the #1 technology ask from the teachers is Chromebooks.

They like iPads up to around grade 2-3. After that, the keyboard makes the kids more productive.

Windows requires too much IT overhead and Mac is too difficult to do business with and expensive.

The real magic of Chrome is the management suite. It’s dirt cheap for .edu and works fantastic. Chrome devices on a cart are reconfigured by the time the cart is rolled into a different classroom.

It’s way too expensive for commercial customers though.

I agree. I'm not saying that Chromebooks aren't a good fit for education. I'm saying that using Chromebooks in schools aren't going to necessarily endear them to students to the point where the next generation is going to prefer either Chromebooks or GSuite when they have a choice at home or when they go to college or start working.

Teachers want Chromebook's for their students. When they have choice, they will use their own laptops for themselves.

Just like no one ever said they want to use Version One for their own personal projects. They would choose something like BaseCamp Personal or Trello.

G is making progress though - automation / building of tools is 100x better on google sheets than in excel.

I can make a g form, output to a g sheet, automatically email if an entry contains certain text, etc. All that takes ~20 minutes, 1 add on, and 0 lines of code. Excel simply cannot compete with that.

Edit - I still use excel all the time for heavy lifting though. They fill different use cases for me.

My initial post may have sounded more negative than I intended regarding the quality of Google Sheets. I have an Office365 subscription and Numbers is free and either could do the simple things I use a spreadsheet for, I still use Sheets.
Have you tried Microsoft Forms and Microsoft Flow? They are the equivalent of your G Forms work flow
Nope, that looks awesome though thank you! My current work doesn't let us use g products
Is there an easy way to set LibreOffice Calc to the Excel keybindings? It might not help for things like macros but at least sheet navigation could be consistent.
What exactly is the story of LO as regards Excel: do they want to emulate Excel's VBA, or do they have a different framework in place(for example using python)?
Does LO support RTD or eg the Bloomberg toolbar?

Can’t replace Excel in many finance roles without those.

is there a list of the missing feature set and is there any group working on the backlog?