Follow Slashdot blog updates by subscribing to our blog RSS feed

 



Forgot your password?
typodupeerror
×
Microsoft IT

Why Microsoft Excel Won't Die (economist.com) 163

The business world's favourite software program enters its 40th year. The Economist: Excel has featured in plenty of workplace blunders -- though its defenders will be quick to blame human error. The financial world is littered with tales of costly spreadsheet errors. Excel has also been blamed for botching gene names in over a third of genomics papers (because it labelled them as dates); underreporting covid-19 cases in England (because it only had a limited number of rows in which to record the results); and disrupting the trial of January 6th rioters in America (because sensitive information was left in hidden cells).

Such snafus have not dented Excel's dominance. Might artificial intelligence (AI) steal its crown? With whizzy new tools powered by the technology promising to make data analysis easier, the familiar grid of numbers and calculations could soon feel outdated. Rather than replacing spreadsheets, though, AI might make them even better. Last month Microsoft introduced an AI assistant for Excel which lets users crunch data using natural-language prompts. Excel, and its faithful, aren't ready to be filtered out just yet.

This discussion has been archived. No new comments can be posted.

Why Microsoft Excel Won't Die

Comments Filter:
  • by Ed Tice ( 3732157 ) on Thursday October 17, 2024 @12:45PM (#64872419)
    As stated in TFS, it's a favorite program of the business world. It's cheap, versatile, and useful. In business, you can't often know a priori what you're going to do with data or even what meta-data you want to collect. Putting the data in Excel is quick and easy. It happens all the time that you realize half way through a task that maybe you want one more piece of data related to an item. Just insert a column. Probably most uses of Excel have specialty programs that could replace it. But one doesn't know at the outset which specialty program that might be. And needs change over time. Put the data in Excel and then transform as needed. It's a solution that works and replacing Excel would be a solution searching for a problem.
    • by i kan reed ( 749298 ) on Thursday October 17, 2024 @12:50PM (#64872453) Homepage Journal

      Software that does a simple job well is valuable. Excel also brings a bunch of dubious upsell bullshit for a million vaguely related use cases, but "Do some math on a bunch of numbers" is what computers were built for.

      • by Ed Tice ( 3732157 ) on Thursday October 17, 2024 @01:10PM (#64872513)
        The *consumer* version of Excel might bring dubious upsell. The corporate version doesn't. Microsoft has an actual sales team for that. The end users aren't the purchasers to nagging them to upgrade wouldn't achieve much.
        • by geekmux ( 1040042 ) on Thursday October 17, 2024 @05:16PM (#64873119)

          The *consumer* version of Excel might bring dubious upsell. The corporate version doesn't. Microsoft has an actual sales team for that. The end users aren't the purchasers to nagging them to upgrade wouldn't achieve much.

          Are you trying to sell me that it was the corporate world that insisted on putting 4,873 font options and an infinite color wheel among the other 90% of features no one truly needs? Because I have yet to find the CFO who gives a shit beyond the number of colors and fonts they can count on one hand. Turns out the corporate world has little need for dubious bullshit too.

      • by Anonymous Coward on Thursday October 17, 2024 @02:27PM (#64872723)
        And for me, "do some math on a bunch of numbers" will remain an Excel job for quite some time. There's a lot of promise in AI, but at least if I make an error in Excel I've only got myself to blame, whereas if I actually want to check the work from submitting the same problem to our current LLM's, I'm doing the work anyway. Not the least of which, being somewhat math-fearless, it can take a lot less time to actually plug the formulas and data in than come up with an explanation of the problem you're solving. Just for amusement, I'm going to type it out now:

        Imagine I have four long-term loans of $A, $B, $C, and $D with daily interest compounding at annual rates E%, F%, G%, and H%, respectively, and no payments due on the loans until they are consolidated. Consolidating those loans in pairs will yield two new consolidated loans with interest rates based on a weighted average of the interest rates of the initial loans, weighted by the loan balances on the day that the consolidation is effective. That weighted average annual rate will be rounded up to the nearest eighth of a percent for the consolidated loan. Given that the interest rate going forward on a consolidated loan will be greater than the effective interest rates on its two initial loans, when in the future should a first consolidated loan be executed to minimize that difference, which two initial loans should first be consolidated, and what is the present value cost attributable to the difference between the interest rates on the other initial loans and a second consolidated loan if the second consolidation occurs on the same day as the first consolidation?

        Took me way longer to express that in English than to just type some formulas into some cells and paste 3650 rows... and if you didn't understand it entirely in spite of my best efforts, why should I think an LLM will?
        • by Targon ( 17348 ) on Thursday October 17, 2024 @02:51PM (#64872795)

          The key problem with AI is to verify the results. Way way way back in the days of grade school, the idea of, "show your work" was done so a teacher could see the thought process for coming up with an answer. For AI, it's around as useful as just pulling the answer out of its ass, it shouldn't be trusted YET.

          • This is and isn't an issue in Excel. The AI assistant provides you the formulas, it doesn't do the calculation itself. The problem is the kind of person who needs to use CoPilot is the kind who doesn't understand the formula, but in this case it is there for you to be verified.

            CoPilot explicitly gives you only advice it can reference from Excel's own help file. For example you can write something like "summarise column A and B and add the sums together" and CoPilot will spit out this:

            Looking at A1:B5, here

        • However a huge amount of use of Excel isn't as a spreadsheet but a database. That was what killed Quattro Pro all those years ago, Borland focused on making it the best spreadsheet possible while MS knew their market and made Excel a less-bad database, because that's what a significant percentage of its users were using it for.

          Think about the last time, or last few times, you saw something set up in a spreadsheet, say on Google Docs. Was it a database of some kind or an actual mathematical spreadsheet?

        • Took me way longer to express that in English than to just type some formulas into some cells

          That's because you know how the formulas work. AI isn't for you in this case. It's for the people who have to open Google every time they do something more complex than add a number and change a cell shade to green.

      • Re: (Score:2, Flamebait)

        by Darinbob ( 1142669 )

        Excel is very complex. I picked up the original Lotus 123 very quickly, when it had no GUI. Because Lotus had a decent and quick help system that was easy to use. It was also very simple. Excel on the other hand has always confused me, its help is confusing, and it keeps adding new and pointless features so that it extremely complex. Anything complex enough to have a Windows style "ribbon" needs to be simplified.

        • by Z00L00K ( 682162 ) on Thursday October 17, 2024 @02:53PM (#64872797) Homepage Journal

          Help in Excel? The best help for me has been Google to figure out some of the more obscure syntaxes.

          One of the absolutely most horrendous things with Excel is that the CSV format is dependent on your localization settings where you have:
          English - '.' for decimals and ',' for field separation.
          Swedish - ',' for decimals and ';' for field separation.

          So when an external program creates a CSV file it's of course in the English format and to read it then you need to pre-process it to get the localized format.

          Then we have all the macros in Excel that are also translated to the local language (WTF were they thinking?). Write something in the US English Excel and it works everywhere, but write something in another language and you are locked into that language. Maybe it has been fixed in later versions of Excel, but at one time it was horrendous. The translated macros are also very obscurely named so I can't use them - and any help you can look up on the net is for English version, which makes it even worse. So if you make macros in Excel in a non-English language you could as well be using Brainf*ck.

          • by i kan reed ( 749298 ) on Thursday October 17, 2024 @03:36PM (#64872905) Homepage Journal

            Here's the lesson:

            You should not use CSV, it does not actually have a standard, and in fact has several competing de-facto standards. It is implicitly a danger to your software's consistency.
            TSV is supported by almost every piece of major software that uses CSV and has a standardized format.

            • by Z00L00K ( 682162 )

              It won't help me when other software that I use creates CSV files.

              I prefer XML instead because that will give even better results when importing.

            • i have been using computers for almost 4 decades, work in IT, have a degree in computer science, been using Excel (for very simple stuff) for over 20 years, and guess what... Never heard of TSV until I read your comment. Never seen anyone else mention it, and worse, never seen any single website use this format. Hence TSV is doomed if even someone like me hasn't heard of it :(

            • by AmiMoJo ( 196126 )

              I wish Open Document Format had taken off, but hardly anyone seems to use it. For spreadsheets the lack of a common macro language is probably holding it back.

          • Help in Excel? The best help for me has been Google to figure out some of the more obscure syntaxes.

            Congratulations! You are now the local computer expert! [xkcd.com]

        • by jmccue ( 834797 ) on Thursday October 17, 2024 @04:15PM (#64872985) Homepage

          A little bit of history, I knew people who worked a lotus in the 90s.

          Microsoft and IBM ensured Lotus that OS/2 will replace Windows/DOS. Lotus, being a small company focused on developing for OS/2 instead of Windows. So, when Windows 95 came out, Lotus had no product for it, but guess who did? That is how Excel took over.

          From what I heard, the OS/2 version was very good. I also think Word Perfect was scammed in the same manner.

          • by HBI ( 10338492 ) on Thursday October 17, 2024 @05:35PM (#64873143)

            To be fair, what happened there had a lot to do with IBM and its behavior. Gates and his cronies were a bunch of dicks, but IBM was rudderless and difficult to work with. One of my first gigs was designing a Prodigy (a joint venture online service basically run by IBM) store in the very early 90s. Armonk was a very interesting place. This being back in the days of everyone wearing a blue suit. Not kidding. The binder full of API details was blue also. Ha.

            Anywho, one detail about OS/2 was that it insisted that the 0,0 point of a window was the bottom right corner, as opposed to top left in Windows. As for why - some functionary in IBM decided that things were going to be that way. Contrary to the rest of the world.

            I think at a certain point, Microsoft probably was on board with OS/2 being successful, but their minds changed over the years from 87 to 90 and particularly with the success of Windows 3.0. So i'm not sure everyone got sold a bill of goods. Microsoft just had multiple avenues to relevance mapped out, and the one that was most successful resulted in near-total dominance.

            • by _merlin ( 160982 ) on Friday October 18, 2024 @12:19AM (#64873571) Homepage Journal

              Anywho, one detail about OS/2 was that it insisted that the 0,0 point of a window was the bottom right corner, as opposed to top left in Windows. As for why - some functionary in IBM decided that things were going to be that way. Contrary to the rest of the world.

              No it wasn't. The origin was at the bottom left, with coordinates increasing up and to the right. NeXTStep and by extension GNUstep, OS X, modern macOS and iOS also work this way. DIB image files (Windows or OS/2 BMP) are usually stored this way, too. So a significant chunk of software does things the "OS/2 way" now.

              As for why, it's the way coordinates work in geometry and on graph paper. Putting the Y origin at the top and increasing downwards was originally done to simplify software on early computers. Conventional CRT displays draw from top to bottom, so video memory was organised the same way, with higher addresses corresponding to lower positions on the screen. Y coordinates were converted to addresses by multiplying with a positive number. IBM and NeXT went back to the drawing board and decided to use conventional geometric coordinates.

              • by HBI ( 10338492 )

                I appreciate the correction, however it didn't take over the world and no one emulated that in a future OS. Because it didn't really make sense on a computer. Today, when you (mostly) aren't writing directly to video memory, it wouldn't matter which way you did it, but it still isn't done.

    • Enterprise (Score:5, Interesting)

      by JBMcB ( 73720 ) on Thursday October 17, 2024 @12:54PM (#64872467)

      The company I work for makes a complicated piece of enterprise software. The tool we made to feed new data into the system is all Excel based. It spits out Excel sheets, you fill them in with your data, and it loads it back in. We use Excel because we asked what people wanted, and every single user said Excel. We have tools to do direct data feeds from other databases, or just about any other data source, but most people use the Excel tool for the initial data load.

      • but most people use the Excel tool for the initial data load

        Yeah that's one of the nice legacy issues. Most people want to use Excel for a data load into a complex tool because prior to implementing the complex tool they stored that data in Excel.

      • > The tool we made to feed new data into the system is all Excel based.

        Dataloader for Salesforce by chance?
      • It came as a huge surprise to me when I learned, upon taking a new job several years ago, that quite a lot of the worlds process automation is based around excel spreadsheets - the process is designed in Excel first using boxes, lines and other things, and then this is imported into the actual process automation tool using an ancient Excel plugin to interpret the "design" and output code.

        It was actually horrifying to see.

      • by Matheus ( 586080 )

        You could easily work for my company and make that exact same statement (Do you??)

        Business people love Excel. Non-Techy people can use Excel. ..and to be clear.. it's just an XLSX file.. We don't even really care if someone is using Excel or OpenOffice or whatever.. it's just a super flexible file format and one we'll have to support for Import and Export probably "forever".

        Hey I created a great UI for that!
        --> Can't I just upload my spreadsheet?
        --> Can you make your UI look like a spreadsheet?

        We supp

    • by ShanghaiBill ( 739463 ) on Thursday October 17, 2024 @01:06PM (#64872501)

      Using Excel to tabulate and display data is fine.

      A few simple macros are also okay.

      But writing complicated macros to encode business processes is dysfunctional.

      It is far better to have separate programs in Python, Java, or VB that pull the data from the spreadsheets, and are subject to code reviews and source control.

      • by Ed Tice ( 3732157 ) on Thursday October 17, 2024 @01:13PM (#64872525)
        If somebody set out to encode complicated business processes in Excel, they should probably seek mental health counseling. But I imagine that, in many cases, Excel started out as a place to hold some data and maybe sum up the rows and columns and then incremental functionality got added over many years.

        In such a case, there are no requirements documents or specifications or anything so replacing Excel becomes difficult since the first step is massive reverse engineering. And, depending on who does the work, the separate programs aren't always better. Although they certainly can be if done by good people using good process.

      • by ClickOnThis ( 137803 ) on Thursday October 17, 2024 @02:38PM (#64872761) Journal

        Using Excel to tabulate and display data is fine.

        A few simple macros are also okay.

        But writing complicated macros to encode business processes is dysfunctional.

        It is far better to have separate programs in Python, Java, or VB that pull the data from the spreadsheets, and are subject to code reviews and source control.

        This, this, this, and this, to the skies!

        Excel is fine for simple stuff. But for many users, Excel is a hammer in search of a nail. Often they know nothing else.

        My biggest beef with Excel (and similar tools) is that it is write-only. Sure, when you're facing a deadline to crunch out a reduction of tabluar data, it can be a time-saver. But good luck six months later, when you look at the same spreadsheet and wonder what you did. "What was that equation I put in this cell over here? What other cells reference it? Arrgh!"

        I had a non-programmer colleague once, who claimed he created a simulation of a physical system we were researching. It was all in an Excel spreadsheet. I respected the guy, and just kept my mouth shut as he presented his results. But inwardly I was screaming "you're doing it wrong!"

        Once I even saw a paper published with a solution to Poisson's Equation [wikipedia.org] that was entirely within a spreadsheet: the boundary conditions were fixed cells and the cells in the interior region were calculated with a difference-kernel equation applied to the neighbors of each cell. It was a bit like a bear riding a bicycle in a circus: not done well, but impressive that it could be done at all.

        Friends don't let friends abuse Excel.

      • That's the big snag with modern spreadsheets, they try to do something extremely complex but without a language or means of managing complexity. Why use $A5 when you can use a variable? Yes, Excel has this stuff, but it's rarely used because that's not what users were trained on. Sometimes you just need some basic programming language like logic When someone gives me an Excel sheet and wants me to modify it without breaking it, I panic. Because it has no comments, no documentation, not reasonably named

      • by hazem ( 472289 ) on Thursday October 17, 2024 @03:33PM (#64872887) Journal

        But writing complicated macros to encode business processes is dysfunctional.

        It is far better to have separate programs in Python, Java, or VB that pull the data from the spreadsheets, and are subject to code reviews and source control.

        Having been on both sides of this, this is a lovely idea in theory that often fails in practice. Having to rely on IT developers to write business-critical software in a timely manner is either cost prohibitive or takes far too long.

        A good example from my own experience was from working at a large footwear & apparel company during the outset of the pandemic. Suddenly there were huge port congestion issues at various ports in Asia causing the system-predicted lead times to suddenly be far too short. The system didn't have the logic to match up specific ports to specific factories to be able to tell which products would be impacted by which ports. The lead-time tables in SAP were only at an "entire country" level. This meant that all the coverage and supply allocation logic coming out of SAP was tragically wrong. Even worse the port congestion was rather fluid and dynamic (e.g. we know it's 2 weeks this week but are predicting orders arriving next week will add 3 weeks), where our install of SAP assumes such lead times are fairly static.

        We couldn't wait a year for IT to come up with something (and certainly didn't have the $10million IT quoted to partially fix the problem). So what did we do? Set up processes to download the data into Excel and augmented it with the factory x port x destination transit time information, and reproduced the reports that were normally coming out of SAP. The non-IT business users had this worked out in just a week and managed to run a multi-billion dollar business this way for nearly 2 years.

        Was it ideal? No. Did it work? Yes? When did IT actually build this capability into SAP? July of 2022. We'll be ready for the next pandemic, I suppose, as long as we can afford to have these changes ported into the next version of SAP we have to upgrade to.

        This is why Excel remains to prevalent and non-IT users build so much logic into their workbooks - it's either that, or do it by hand.

        • This is why Excel remains to prevalent and non-IT users build so much logic into their workbooks - it's either that, or do it by hand.

          Those aren't the only options.

          A better alternative is for the non-IT people to learn Python.

      • by cusco ( 717999 )

        I've encountered gigabyte-plus Excel spreadsheets full of spaghetti code encoding critical business functions before (I ran away as fast as I could!) If something is so complex that it needs more than half a dozen macros then it should be in a database. Unfortunately the DB interfaces are uniformly horrible and executives refuse to learn how to use them, so we're going to be stuck with this cruft for years to come.

        One of the unsung heroes of the Y2K effort was the team which dealt with the electricity tra

      • What I find is that inevitably someone is going to find the lookup functions and the next thing you know they've created a fragile badly implemented database.

        I have actually come to loathe Excel

    • It's a solution that works and replacing Excel would be a solution searching for a problem.

      That's actually not true at all. In many cases Excel does the bare minimum but isn't sufficient. Excel is a great little tool for storing data in a bunch of sheets. But in most situations where a project attempts to replace Excel it's because Excel is too limited in what it does, or the dataset has grown unmanageable, or management has asked to report some new data that can't be extracted from Excel alone, or you need to interface with other databases in ways Excel doesn't support.

      There are many problems wi

      • Maybe I wasn't specific enough. Sorry for that. The *wholesale* replacing of Excel is a solution looking for a problem. Replacing some *individual* Excel spreadsheets is generally a positive sign. What often happens is that something starts in Excel but, as the company grows due to success, Excel is no longer good enough. And those Excel spreadsheets are migrated to something better. Maybe a customer problem. Maybe an ERP system.

        I've never heard of an "Excel replacement project," but it's certainl

    • Probably a journo who needed to put AI into a sentence?
      Other than that nobody at all, who actually uses it.
      Never a kind word about Microsoft leaves my lips, but SPREADSHEETS are a remarkable tool that is usable by both amateurs and experts for many many different purposes. It is unusually easy to start using and is extensible pretty much as far as you can imagine, as I have seen in my professional life and as many others below point out.... you can do ridiculously complex things with a spreadsheet if you pu
    • by tlhIngan ( 30335 )

      There's plenty of reasons for Excel to die, and plenty of reasons for it not to.

      Spreadsheets existed long before computers - they usually were chalkboards of numbers that people edited and others had to recalculate. When VisiCalc came out, this was a breakthrough in that you can update numbers and have the calculations performed automatically.

      For things like data analysis, they do things in a quick and easy format. And heck, putting data in a quick tabular format is often a convenient way to organize stuff.

    • Multi user database. Spreadsheets are incredibly useful but it can be an absolute freaking shit show when people try to make them do things they're not supposed to do. And God help you when somebody discovers VBA and the ability to add buttons and user interface components. I've seen entire multi-user applications built on Excel and nothing else and they work about as well as you would expect
    • Excel is great. It's needed.
      But I will never understand why it so horribly mangles the data I copy pasted from SQL Server, so that I have to go into each column, change formatting, copy the data again, to fix it.
      Just default to basic text format instead of trying to convert things into dates or exponential number representations or whatever you dumbass piece of shit software!

  • It raised the self-esteem of middle managers all over the world.
    • I know someone who had foisted upon them a resource planning tool for managing thousands of employees across multiple sites and groups, completely written in Excel, VBS macros, and the PivotTable extension. You ran the "master sheet" script and it would pull from other workbooks, creating a massive report in the main workbook. It was basically a complex relational database built using spreadsheets, with no constraint or relationship management at all.

      Finally their IT department built something to replace it

      • As Abraham Maslow put it, "if all you have is a hammer, everything looks like a nail."

        Another, more interesting way of putting it is, "We tend to formulate our problems in such a way as to make it seem that the solutions to those problems demand precisely what we already happen to have at hand." - Abraham Kaplan, 1964
      • I think it was done in SharePoint, and it wasn't as good.

        Obviously. SharePoint isn't useful for anything. Avoid it like the plague.*

        * Having seen many people's reactions to the last plague, we might to reconsider using this phrase.
      • I think it was done in SharePoint, and it wasn't as good.

        No need to recite a tautology.

    • I assume you're talking about VBA, right?

      VBA has been disabled in Excel for years already. You have to go through a bunch of security hoops to enable it, overriding warnings that will make any IT department scream in horror.

      VBA macros haven't been an issue since a few years after the "I love you" virus made them an existential security risk.

  • by MilliMicro ( 6251190 ) on Thursday October 17, 2024 @12:46PM (#64872427)

    Might artificial intelligence (AI) steal its crown?

    Oh sure, take the bit of software which causes errors because of obscure gotchas, and give it AI so it can blindly blunder into those gotchas without any human input at all. I can't possibly understand why AI PCs aren't flying off the shelves.

    • Ssssshhhh!!! dont warn them of this huge risk, just make some popcorn and wait for the trainwreck it will cause in accounting and IT sectors
  • by Pseudonymous Powers ( 4097097 ) on Thursday October 17, 2024 @12:49PM (#64872443)
    A significant fraction of the spreadsheets I am subjected to in the course of my work don't even use a single formula. Many managers just like having a grid to put things into.
    • by ip_vjl ( 410654 ) on Thursday October 17, 2024 @01:49PM (#64872613) Homepage

      To be fair, if it's anything like the ones I produce, it's not that the data doesn't require calculations, it's that the spreadsheet is just the final output format from another process that does the heavy lifting.

      The real work is in the database, the XLS file just is an easy way to distribute the resulting data in a way that nontechnical users can filter, sort, and add fields to. But any of the calculations were done behind the scenes creating the file and aren't ever part of the output sheet.

    • I call that "paper logic for Excel". They don't understand the functionality and just want things to look organized and pretty, but in how it would look if they printed it on paper, but then you can't use Excel's best features with a well designed "paper logic" spreadsheet.
  • by Chris Mattern ( 191822 ) on Thursday October 17, 2024 @12:50PM (#64872449)

    "No matter what the actual requirements document says, what your users really want is for you to implement Excel for them."

  • by Kiliani ( 816330 ) on Thursday October 17, 2024 @01:05PM (#64872497)

    As much as I dislike Microsoft, Excel has its place. I use it for planning purposes, mainly budgets, and the occasional data recording (temporary recording before transferring data to processing software). Also works for mocking up rough schedule drafts, drafting a simple table to be ingested elsewhere, tracking tasks when a multirow, multicolumn table is needed, or for storing lookup tables to be used in software elsewhere.

    So plenty of uses, especially if much math is not really needed. But the moment serious calculations come into play ... yeah, not so much. Unless you are very careful. Returns diminish quickly and other tools are just more appropriate / better / faster / more reliable.

    In the end, not surprising. Spreadsheets *do* have their place.

    • This is accurate. It's not Excel. It's spreadsheets. Spreadsheets are highly customizable to a unique situation. Excel is dominant because it's the best spreadsheet tool on the market, the next closest competitor is maybe Google sheets which has dubious functionality IMO.

      Every other spreadsheet tool is a niche case useful tool, but Excel is the top.

  • by apparently ( 756613 ) on Thursday October 17, 2024 @01:26PM (#64872551)
    Yes, the technology that literally just makes stuff up is definitely what will ensure that spreadsheets contain accurate data.

    Such snafus have not dented Excel's dominance. Might artificial intelligence (AI) steal its crown? With whizzy new tools powered by the technology promising to make data analysis easier, the familiar grid of numbers and calculations could soon feel outdated.

    You literally can't trust the output of "A.I." to be correct. Ever

    • This. Times 1000.

      Taking a small business level application and adding "AI" to it will only multiply the problems.

      The problem with Excel isn't Excel. When used by home users and small businesses as intended, it is fine.

      But like most of you I have seen large enterprises abuse the shit out of it for stuff it was never intended to do. Don't blame the tool. Blame the carpenter.

      There are several enterprise capable options available in the real world for 10+ years which are multi user, multi dimensional, have

    • by Anne Thwacks ( 531696 ) on Thursday October 17, 2024 @01:51PM (#64872629)
      You literally can't trust the output of "A.I." to be correct. Ever

      Most people don't want "correct" they want the answer they are arguing for.

    • You literally can't trust the output of "A.I." to be correct. Ever

      True, but you typically do not use machine learning in circumstances where you need it to be 100% correct all the time. For example, in particle physics we use machine learning algorithms to separate signal events from background events. The standard selection criteria that they replace are never 100% accurate in separating the two so you always have to measure the signal efficiency and background rate remaining.

      What you can do is measure these parameters for a trained ML algorithm and then trust that i

      • by Targon ( 17348 )

        That phrase, "Machine learning" says it all. It needs to be trained before it will know how to do something well. Move to the next step, the AI knows how to do a limited number of things, so will be good for automation.

        • The current usage of "AI" is synonymous with "machine learning". We do not have any real "AI", yet at least, and I use "machine learning" because that is a more accurate description of all the current systems, including ChatGPT and the like.
  • by iggymanz ( 596061 ) on Thursday October 17, 2024 @01:27PM (#64872555)

    The idiots at England's PHE (Public Health England agency) used an archaic file format, XLS from 1987, that could only take 65K rows. Excel can have over a million rows on a sheet.

    • Over a million? Seriously? Then Excel hasn't learned its lesson and is repeating the same mistake over and over again. We just have to wait for the next disaster, when someone will try to paste two million rows in a new format spreadsheet.

      But you're right that idiot users are to blame... but for using Excel in the first place, rather than for incorrect use of Excel (because Excel is laughably easy to be used incorrectly). Their bosses that allowed using Excel for serious data handling are no less to blame.

  • Like anything that "won't die", Excel has value and allows a user to create results/work faster and cheaper than not using it. Same goes for Word, computer mice, keyboards, color monitors, the Internet, photocopiers, and tons of other office tools.

    • by gtall ( 79522 )

      You mean it allows any yahoo to bang together something in Excel and foist it on people who have no choice but to use his/her pile of rat droppings. The percentage of people who can use Excel well pales in comparison to the legions to think they can but are blindly misguided.

  • by MpVpRb ( 1423381 ) on Thursday October 17, 2024 @01:41PM (#64872581)

    ... it sucks as a programming language
    There are no built in tools for visualizing and managing the complexity that can be created, and it's possible to create an insane amount of complexity with it

    • by thegarbz ( 1787294 ) on Thursday October 17, 2024 @02:10PM (#64872683)

      There are no built in tools for visualizing and managing the complexity

      Excel literally has a Python interpreter for complex data visualisations built into it. But then honestly I struggle to come up with a scenario where you need to create and visualise complexity in Excel. The whole point of Excel is literally the opposite, take something complex, simplify it, and then display the simple result.

      Despite it's capabilities (like the aforementioned Python interpreter) if you're visualising complexity in Excel you're using the wrong tool for the job.

  • by ThurstonMoore ( 605470 ) on Thursday October 17, 2024 @01:47PM (#64872607)

    Anyone who's had to deal with Google Shits knows why Excel wont die.

    • Anyone who's had to deal with Google Shits knows why Excel wont die.

      Yeah, I've been working on a document I keep in Google Docs for ease of portability. I created a simple little graph for a scenario based on some modified sine waves in Excel. I wanted to keep the data supporting the graph in case I want to change it in the future, so I though just upload it next to my doc. Google did many bad things to my simple little graph that was a much greater pain to fix than in Excel.

  • by Tony Isaac ( 1301187 ) on Thursday October 17, 2024 @02:05PM (#64872663) Homepage

    These days, it doesn't matter whether you actually have Excel. You can use Google Sheets, or LibreOffice, or any of a number of unknown brands that you might encounter on various websites. They all work for what most people need, and they all produce interchangeable files. It's as ubiquitous as Notepad.

  • by thegarbz ( 1787294 ) on Thursday October 17, 2024 @02:07PM (#64872673)

    Why would Excel die? It's not like literally any of the alternatives would make the problems disappear. Hidden cells, spreadsheet errors, that is a function of every spreadsheet. Misclassifying data happens all the time in databases. As for the COVID cases hitting the cell limit, they weren't using "Excel", they were using a woefully old and outdated copy of Excel and saving files in the previous file format, they literally wouldn't have had that problem if they were up to date.

    Excel is a tool. A tool can be used for good, or for evil. Just because a Philips head screwdriver isn't very good at driving nails into the wall doesn't mean you shouldn't use it with Philips head screws.

    • Excel also does MONEY incorrectly and every accountant has just gone along with it not knowing except maybe some really old ones long ago who remember banker's rounding and didn't know what a float was. they still don't know what a float is or why you don't store money in one. To think of all the tons of money just floating around ... except for the few who collect it over time... while the rest just don't bother to deal with it.

      Money is integer math; in pennies which BTW, almost everything on earth goes d

      • Excel also does MONEY incorrectly

        Like all the nonsense in the summary, this is a USER ERROR, not an application issue.

      • Excel also does MONEY incorrectly and every accountant has just gone along with it not knowing except maybe some really old ones long ago who remember banker's rounding and didn't know what a float was.

        You not understanding what the "Precision as Displayed" option means doesn't mean Excel is doing anything wrong. It means you're using the tool incorrectly. If you're calculating large currency tables you should google this option and learn how to use Excel correctly. It's right there for you, even when you deal with *takes breath so he can shout* MONEY!

  • by dmay34 ( 6770232 ) on Thursday October 17, 2024 @02:36PM (#64872755)

    Apple's Numbers spreadsheet program is awesome. If you've ever used it, it's way different than any other spreadsheet program and is geared more toward data presentation. But it's SO fantastic for data presentation! I wish everyone else's spreadsheets worked more like numbers, or had similar functionality.

    • I can't tell if you're being sarcastic... Numbers is still deficient compared to Excel in many presentation features. Numbers' graphing is sub-par, especially for math/engineering; specifying series for plotting is also painful. You can't do text rotation in cells. You can't have multiple formats (bold, regular, and italic) in a single cell.

      Basic stuff like that.

      • by dmay34 ( 6770232 )

        Yes, there are a lot of reasons I don't actually use Numbers.

        But the ability to make tables and put them wherever you like on a page and have them all talk to each other is awesome!

  • Huh? (Score:4, Insightful)

    by The MAZZTer ( 911996 ) <megazzt.gmail@com> on Thursday October 17, 2024 @02:57PM (#64872805) Homepage
    All the "problems" listed in the summary would occur regardless of which software you used. It's not Excel's fault, the constant IS human error. Not enough rows for results? Why were they using Excel for that, clearly it was an inappropriate use of Excel and they should have used something better purposed for tracking large amounts of data. Hidden cells? Someone hid them, Excel was only following directions. The date thing? Excel is just assuming people don't know how to properly use Excel . Which is understandable. Also you can override this behavior, which just shows the scientists do not know how to properly use Excel to do so.
  • Oh please Microsoft add the support numbers in SI engineering units.
    I just want to see 120p instead of 1.2E-07.
    At least CCalc has done this since forever.

  • by kick6 ( 1081615 ) on Thursday October 17, 2024 @03:12PM (#64872841) Homepage

    Billion dollar business decisions are made on the back of an 80MB excel "net asset value model" that takes minutes to recalculate because it's a labyrinth of references, built by sleep a deprived 23 year old analyst at 2am, overseen by a sleep deprived 25 year old associate, with one final check by a distracted Vice President.

    Of that entire stack...the software is probably the LEAST concerning part.

  • by Ogive17 ( 691899 ) on Thursday October 17, 2024 @04:19PM (#64873001)
    Can't blame the tool for a user error. Every piece of business software is going to provide an incorrect result if the user messes up the input.

    I use OpenOffice Calc at home since I really only track expenses. Even though there are many similar functions to Excel, it's just clunky to use and some simple Excel functions appear much more difficult in Calc (though I can admit some of that could be chalked up to my unfamiliarity).

    My only complaint with Excel today is they keep adding "feature" or enhancing the operations that just add bloat.
  • It's interesting that there's this intuition that a piece of common software over time will become a commodity: cheap and you can't make money with it. Somehow a package can remain a cash cow for a very long time. How come? You can keep adding features but if people decide they don't need those features they aren't going to pay for them.
    There is the lockin from file compatibility. "Yeah but to really have reliable formatting you best use Word and not some cheap alternative" .

  • And that even if AI worked and was useful, probably especially if AI worked and was useful. It's not fucking needed. It's like buying something you will never need because it's on sale. Just fucking stop already.

Never test for an error condition you don't know how to handle. -- Steinbach

Working...