Hacker News Reader

Comment by hylaride | Hacker News Reader

I don't think people realize how much of the world runs off of excel, especially the financial world. Excel still "supports" buggy issues from versions in the 1990s as to not break financial models.

IBM mainframes do the same thing with COBOL because modernizing core code means having to notify and test all the downstream users that have their own code to deal with the legacy issues in the current stack. I was once explained by a retired COBOL programmer (who spends his summers contracting for maintenance) how this tangled mess of thousands of banks feeds in all the way to the US central bank to determine interest rates - in some extreme cases, IBM still has to emulate hardware bugs as to not change expected behaviour. This is not something you just migrate away from or fuck around with. There's a reason IBM still sells mainframes that runs code from the 1960s.

View on HN

> I don't think people realize how much of the world runs off of excel, especially the financial world.

Most people working in software understand this as a huge chunk of software projects are "let's get these people off of their spreadsheets!"

I know what you mean, but now you're talking about a very niche market. Finance, which uses Excel or spreadsheets since 90s. You can't replace these, I agree.

But for other uses of Excel, I can say it can be easily exchanged. This is why I used 99.99% percent, but not more nines. The remaining "three nines" are the niches you're talking about.

It's like how we say "Nah, it's a small set of matrices. just 2, 3000x3000, double precision floating point matrices, and a couple of vectors and some multiplication (i.e. in a range of a couple of billions)".

It sounds like a niche market, but every company in the world has a finance department that plugs everything from sales forecasts, various inputs (goods, employment, supply chain components, etc), to things we don't even think of that flow through it. At some point these companies are talking with other companies and a shocking amount of data interchange is done by emailing excel spreadsheets around (or CSVs exported from excel).

I forget about it myself as I've worked for modern tech companies since 2008, but outside of that world is a very different place where old ways of doing things stick around. Anecdotally, about 2 years ago I got a cold email from a co-op job that I had in 2002 for a manufacturing company that I wrote a perl script for asking for help modernizing it. The perl script consolidated raw EDI(1) files coming into some ancient HP-UX machine, then exported it...to an excel file, MS Access DB, and a SAP DB so that the accountants and the manufacturers could simultaneously get work processing the order. When I started people were inputting these all manually.

1: https://en.wikipedia.org/wiki/Electronic_data_interchange

I recently switched from Excel. I'm not a FOSS ideologue, but I found that Calc does more of the things I need with less of the distraction, more customisability, and less irritation with bugs and breaking updates.

It's true I'm not a spreadsheet power user - but not many people are.

The trouble is that excel really does have a disproportionate share of all the tiny applications that keep the world running, so it's very very hard to move.

I remember back in the 00s Excel could support millions of rows on an office PC used in a bank department to check approvals for loans. I don't want to know what perversions they're doing now but at some point all that needs to get in normal DBs.

Excel has native DB connectivity for quite some time. Maybe they were using that, actually?

Libreoffice Calc also supports the same thing over ODBC AFAIK.

Damn, you might be right. I even made a tutorial on Excel ODBC for our sales department recently and completely forgot it's ancient tech. But even then, why in the hell would you query that much shit outside of management diagrams...

Reply on news.ycombinator.com