His "arguments" are at the "This Is Gatekeeping" paragraph. It boils down to "spreadsheets are too fragile to be the foundation of any business-critical data pipeline"
I must plead before HN that I actually didn't post this myself because I wasn't making a serious case, and was mostly just having a bit of fun. I don't want business users to feel bad about their spreadsheet use because it does empower them, but they certainly drive me crazy at times. However, for some actual points, discussing spreadsheets specifically and ignoring SharePoint etc:
- As another pointed out, the sheets tend to have no constraints on what users enter
- Users will tend to begin performing ad-hoc queries inside the spreadsheet, making reading it a nightmare later on
- If you look away from three seconds, the data will be pivoted into a format that makes it super hard to move to a database later on
- People will start to keep multiple versions of the sheet around like 'analysis', 'analysis_final', and 'analysis_FINAL', scatter them around a network drive, and trust God to sort them out later
- Probably a million more things that I can't think up ad hoc
At work, I do relentlessly advocate for us not to use spreadsheets, though I certainly don't threaten people in this way. And that's because we have the resources to do things better, and instead of empowering people the right way, we get them to do everything in Excel instead of leveraging the real power of the appropriate technology. There's probably an extensive place for Excel, when wielded responsibly or in contexts where there's no way at all that you can afford the staff to manage a more robust solution.
And, most importantly, I was hoping the judge would be lenient if my lawyer could reasonably argue that I warned the victims.
Although I don't agree 100% with this entry specifically, it is particularly refreshing to read something unfiltered, witty written and yet unafraid to meet controversy. (which don't often go hand in hand lately)
As much as I hate how unstructured spreadsheets can be (and almost always are), I have successfully integrated spreadsheets into business processes by developing a schema for the spreadsheet and enforcing it. I have web apps which use the Apache POI library to parse the spreadsheet. They kick it back to the user if anything is not where it is expected to be. Less technical end users can then use Excel as their UI for entering data into forms. It works great for some use cases.
You are a person of iron discipline, and I respect it. This was almost done at my current workplace, but the spreadsheets were not adequately formatted/enforced early on, and there wasn't political will to insist users re-do work if a newer version had been produced, so it fell to the engineers.
I give you, and only you, a pass to use these in the future.
Having only barely touched VBA in a single spreadsheet for a few weeks at a $B company I can relate. And don't remind me of the poor colleague tasked with re-implementing a simple searchable HTML table in SharePoint, who spent at least a full year on what would've been a couple of weeks in a reasonable CMS.
Well the reality is, excel is not going anywhere anytime soon.
It’s one of the reasons why MS is so big.
It’s so far entrenched in the business world, so even if Im also gonna kill myself the next time i have to do a fucking VLOOKUP and it craps out on me, I will have to do it because Johnny from accounting surely cant deal with SQL or anything else (pandas, gasp!).
Even if Microsoft would fuck up Windows badly and looses market share to other OSes, people will still be using Excel for the next one hundred years.
What you probably want is a platform that will isolate spreadsheets but enable them to be part of the process. Shameless plug: our SaaS planning platform [1] enables collaborative Excel-like spreadsheets, full data science analytics and actual Excel as part of the workflow; but you control the workflow completely.
I am going to look at this, and if it is cool I may even bring it to the attention of the capricious gods capable of buying software at my company, but seeing Excel and Data Science in the same sentence, while individually concerning, fills me with a deep sense of foreboding.
Well, I get the concern. Our platform keeps a business process graph in an actual rule-based language and enforces that graph continuously (literally 24/7). Some of the process steps in that graph can be executed in proper Python --in parallel, on the cloud, for TBs of data-- and some can even be executed in Excel (locally, for specified and controlled scope, hopefully for MBs of data). Excel is in the picture but in a controlled sandbox that can be avoided/automated away when the business team is ready.
While there is little more frustrating than a spreadhsheet pretending it's a database this seems like it was written by a first year IT associate who doesn't really understand either side of this cycle.
Their advice boils down to do something stupid with a higher support burden rather than understanding when to use each tool in the arsenal.
While I'm deeply sympathetic the author should be discussing their issues with a therapist rather than spreading this on the internet as of there's some real basis for it.
If software developers could produce user maintainable software at anywhere near the pace the users can build the spreadsheets, the author might have a legitimate grievance.
Instead, the software developers are complaining about people cutting their own hair, essentially.
I actually wrote a little disclaimer at the bottom, because this is largely true (I have some quibbles, but I'm 90% in agreement), then decided that backing down at the end of what is obviously meant to be a funny piece was terrible.
Now, I was looking for actual arguments on why one should not use spreadsheets, and I haven't found anything really convincing in this post.