Every visual workflow tool is just Excel for developers who gave up
medium.com
medium.com
> You know what happens with Excel. Karen from accounting builds a “simple” spreadsheet to track expenses. Six months later, it’s a 47-sheet monster with circular references and VLOOKUP formulas that would make a mathematician weep. The company depends on it, nobody understands it...
It sounds like Karen did a valuable service to the company. She combined a technical skill set with her domain knowledge to create a system that was so successful that the company now depends on it. Cleaning up the technical debt seems like a task that's well worth the cost.
> ...And when it breaks (not if, when), guess who gets called to fix it?
I'm not sure, honestly. It depends on whether you want the fix being done by a resource you consider to be a cost center or a value center. The former will do the cleanup job for bottom dollar. The latter would team up with Karen to amplify the project's impact while cleaning it up.
For these reasons I think your analogy is not effective. I don't disagree with your thesis, though; I prefer code in almost all cases. But that's just me. I know for certain, though, that if I had stock options in that company in this hypothetical scenario I'd rather keep Karen around than whoever fixes the VLOOKUP syntax. And if visual tools are what empowers Karen... well, there you go.
The technical debt here was solved by creating a complex Excel worksheet. The sheet is the solution.
For small companies where these Excel monsters get created, it is the #1 best way (read - cheapest) to solve technical debt, which, before Excel, was probably a bunch of arcane manual processes that took 5x as long with worse accuracy.
With Excel you get no-code immediate UI feedback in the cells.
Most people will use an Excel sheet as an array or a dataframe.
This is constantly repeated truisms, and yet I don't use it and have no plans using it. I'm certain I'm not alone. Excel is counterintuitive and not 'consumable' as the opposite to SQLite. Surely I understand it not the same. Also I suspect most of the Excel users use it like CVS editor.
"Karen" can literally write programs with little training, there's memory, there's functions, there's references, excel is the best.
I do not particularly like excel, I find it interesting but often obtuse and limited in annoying ways and I agree that for an established workflow using "proper" code is generally better. But imho where Excel excels is in dynamic exploration and supporting decision making, where the type of analysis and visualisations you might want to do can change radically and quickly
The immediate feedback and visual layout make it the definition of "killer app." Jupiter notebooks has been trying to replicate the magic for years. Entire businesses and industries are built around selling products to successfully replicate what was originally living in a spreadsheet somewhere.
Too easy to refute. Treating "Excel" as actually "spreadsheet processing," the software has been so consistently and hugely popular that everything from early PDAs [0] to Google Sheets / iCloud Numbers has supported it. Every type of personal computing device has it, from graphic calculators [1] to laptops to tablets to PDAs to many cell phones.
> I suspect most of the Excel users use it like CVS editor.
Like some classes of addictive chemical compounds, you may start with this but quickly find it is the gateway to other habits.
[0] https://en.wikipedia.org/wiki/HP_95LX
[1] https://education.ti.com/en/software/details/en/139B977E62CB...
I feel like I'm hearing "Hundreds of people who don't like Elvis can't be wrong". And I don't even like Elvis!
And, what's git? A spreadsheet can be committed to git.
Version control for spreadsheets? It'd be awesome to have on its own, but it's also not hard to add it on top of spreadsheets' existing features.
That's exactly my point.
It's always interesting to see where developers don't build obvious solutions into existing tool.
You're very right that it's not hard, but to me, the fact it hasn't been done, is interesting thing from a sociological perspective.
I've seen many code editors with amazing git ergonomics built in, never for spreadsheets.
I'm actually working on text only spreadsheet markup that is human intuitive and building interfaces to go with it.
This is not the first time I see developers berating non-technical users, and I like the thesis that they will function as cost-centers that will do a cleanup job instead of teaming up with the user. It's like they will attempt to "do the work" in a binary fashion, making the "correct" choices, often focused on technical factors, like which library to use or what algorithm, instead of focusing on the business domain.
I also see this attitude when facing technical issues or bad code as if it wasn't our job to deal with bad code (and as if non-bad code existed)
>I'm not sure, honestly. It depends on whether you want the fix being done by a resource you consider to be a cost center or a value center. The former will do the cleanup job for bottom dollar. The latter would team up with Karen to amplify the project's impact while cleaning it up.
You're right that the answer is "it depends." In the world of risk management and audit, what Karen has created is called a "user-developed application." Her employer should give her a big pat on the back, and then take a step back to decide 1) if this Excel spreadsheet is mission critical, 2) if it is, who will maintain it, and 3) whether its importance justifies building a formal IT solution to support the business function (either in-house or using COTS software).
I agree that Karen did nothing wrong! She found a process that could be improved or automated, and was able to use technology to do it. If someone is caught off guard when the Excel file goes kaput, then that's not Karen's problem nor is it her fault; it's her employer's responsibility to pay attention to their own business requirements.
So is Photoshop and Google Maps - the point of the article isn't bashing on Excel, it's about choosing the wrong tool for a job.
And that avocado slicer vs. kitchen knife analogy is perfect, it captures the nuance so well. Some tools are brilliantly specialized, while others thrive on flexibility. Maybe the real takeaway is that knowing when to reach for each is the mark of a savvy developer.
Thanks again for the great discussion, this is why I love sharing on Hackernews, nowhere else has such a density of curious, sharp-minded people who push ideas further.
I was a bit confused by the post because it seems to be written about a persona that I haven't personally come across: professional developers who choose visual workflow tools as a primary solution over building with code the "right way".
I'm still a bit confused because you make broad statements about these tools and their users while actually you seem to be focused on this specific subset of users. e.g. when you talk about why these tools are successful, was your intent to limit the scope of your comments about the success of this category to just professional developers?
In a separate comment, I criticized the post before I fully realized you're specifically talking about developers, and not the typical audience for these products. I'm trying to understand if that's the lens through which the entire post should be interpreted. If so, making that more clear would strengthen the post I think.
I'm still a bit skeptical about the existence of said persona, but happy to be shown I'm wrong.
That said, there are a few points against this as a complaint. First, is the incredibly important point that many many tools are people rediscovering pivot tables. There was a fun rant a while back about "your startup is just a pivot table." Hilarious read.
After that, visual workflows are clearly easier for people to understand. Just look at the directions you get with any "self assembled" furniture. Some of that, I'm sure, is to avoid having to translate a lot? Hard not to argue that it is still probably the more effective way to communicate things.
My final caveat is that the symbolic nature of program code is one that is largely lost on people. Specifically, people seem to think the software is independent of the execution environment that is necessary to understand in the language they are using.
I'd love to read this, but can't seem to find it. Any idea where I could find it at?
Karen from accounting made an entire app even though she's not a "software developer" by trade and people think that's a bad thing?
The real super power to excel is you can see the data as it flows through the cells. Whenever I see a visual programming tool that is just a way to write a regular script by linking up a DAG I can't help but feel like they missed the point. Motivated non-programmers have been using COUNTIF, VLOOKUP and such since forever. They're not incapable of writing a textual program 1 statement at a time, it's just that most programming environments don't facilitate it.
(Not picking on developers in particular. Similar story applies to salespeople, lawyers, technicians, scientists, accountants, or anybody else.)
But I think the author conflates 1) engineering practices
2) Programming
3) Expressing domain knowledge using tables, charts and numbers
Most excel users do 3) They are certainly not programmers, and the do not follow engineering practices we associated with programming: modularity, hiearchical or functional re-usability, testability, etc
Excel creators do not reproduce bug reports, do not measure their success by the number of bugs they fixed, they do not do SDLC...
These are not particularly fault of excel. But as soon as there is a recognition that 'oh that excel workbook needs to be managed by programmers' -- then after a few month programmers realize that Excel and VBA macros are subpar programming environment, and will just argue that the way that the author does -- that it does not make sense to maintain a complex, business critical code within Excel.
So they will re-write it in Python or something else.
---
I think Ex cell's success in the business world is precisely because it allows non-programmers to bypass the whole SDLC and express their domain knowledge in a computational system.
As programmers, we want to replace Excel with something that
a) allows business users to do what they want to do, without SDLC
b) when their work reaches a certain level of complexity, we want the tool to allow us to gradually transition the work the domain users have done into an engineered piece of software with minimal effort, free of cost, easy to debug.
I do not think such as system exists yet, but that would be a killer app
I've seen business-side product owners complain "why can't you just make it do XYZ this week, Excel can do that" without realizing the application is being built by like three people and if they want the feature roadmap to be on time -- that they themselves planned months ago -- doing something seemingly trivial might be a non-trivial refactor. Trying to keep them focused on the long term is like dealing with a toddler's attention span.
So then they get impatient and the business goes out and buys Salesforce, because it "does everything for cheap", and then they quickly realize if you want to do some non-trivial thing -- which their custom application of course needs to do -- they're shit out of luck or have to buy Marc Benioff a new island.
Otherwise it’s a decent tool, but the fact that you write your code in a cell that is so tied to other cells with non-useful names that it’s impractical to reuse is a reason why it sucks.
Whereas many of these visual worksflow tools at least can export to json and be manipulated programatically in a useful way.
I mean if you write a really long formula in a cell, it's not really practical to use that formula in another workbook that doesn't have the exact same structure.
If you compare that to say a Jupyter notebook which is cell based, you can at least reuse your function in another notebook or cell within any other jupyter notebook rapidly and quickly if you wrote it correctly. Excel not so much.
In every project the business users have no interest in “managing” business workflows, so the developers that built the tool end up having to use the tool for them, implementing their business use cases.
Developers are much more comfortable with code. You cannot easily unit test visual workflow designers. You cannot easily version or put visual workflow designs into source control.
Whenever I see one of these projects I run screaming in the opposite direction.
Locode in general is an anti pattern. I’m specifically looking at Microsoft Dynamics and Power Apps, aka. Microsoft Access Online
I would like to be able to upvote this post 100 times.
At that point, you’re stuck using a tool that’s harder to debug, test, or version and still coding on the side. Totally with you on this being an anti-pattern in most serious use cases.
The characterization of visual workflow tools is also an incidental affair: there are commercial tools, and there are open source tools. Just depends on the kind of support one wants, and many devs would love some decent support when some component breaks or something's missing, hence the more likely adoption of a commercial solution.
And in the end it's also just the next evolution of software development tools that's being sought. We moved from flipping switches, to boxes of punch cards, to Assembly, to C-like, to Python-like. We're looking for the next form, and 2-D node-like programming does seem an improvement over 1-D text, as text was an improvement over reams of punched paper. And there will always be those who continue to enjoy doing Assembly, even though the majority moved on to higher level languages, which will also be the case for the current and next level tools.
The value prop of these "Low-Code No-Code" platforms has absolutely 0 to do with the actual "user/developer" experience being better/easier/more powerful and everything to do with the orchestration, inheritability, visibility, management, and security capabilities that these platforms provide. If I have 5 artisanal, bespoke Excel workbooks that my "developer" accountant runs locally to complete a part of a critical business process (and they invariably do) then I have 5 ways my business can come to a grinding halt when any number of things happens to that accountant or their computer. I would take 50 less powerful, rickety RPA workflows that I can at least see in a control room over those 5 Excel workbooks any day of the week.
The reason why my startup uses Zapier isn't because we prefer to use no-code to orchestrate this specific workflow. It's because it's faster than building out all the webhooks, routers, integrations, tables, etc necessary to make this workflow work, stuff that Zapier already natively supports
When a tool crosses from (b) into (a), we'd have to acknowledge that too. (Or ignore it and invite a blizzard of problems, which are now inexplicable because "it was done".)
Every thinking person already in fact does differentiate between (a) and (b) in their own work. So admitting it out loud and just talking about it may not be the apocalypse that many expect. But ymmv.
I don't think the author understands why these tools exist or why people find them valuable, and there are a number of major issues with their position.
> Excel and visual workflow tools are fundamentally the same: point-and-click interfaces that let people build complex logic without understanding what they’re actually building. Excel uses cells and formulas, visual workflows use boxes and connectors, but the principle is identical. Both promise to make hard things easy.
While I sort of understand the point they're trying to make, I think it's problematic to call these "fundamentally the same" but then compare them on abstract characteristics. By this logic, a car and a bus are fundamentally the same because they both promise to get us from Point A to Point B. This of course misses out on a myriad of reasons cars and busses are quite different in practice.
> Visual workflow tools aren’t succeeding because they’re better, they’re succeeding because we’ve gotten lazy and scared.
These tools are succeeding because:
- They enable teams and individuals to build things they otherwise couldn't without involving central IT or some dev team
- They provide structure and promise to solve a host of operational/management issues associated with keeping an automation running in perpetuity
- They can often be charged to an expense card vs. requiring headcount, existing resource allocation, new budget, etc.
Is the result also a monstrosity? Possibly. But It's a monstrosity that is making something happen that otherwise wouldn't be happening.
And it's possible that the monstrosity will eventually need to be adopted/fixed by a real dev team, but again, that's a team that would never have gotten involved to begin with if someone hadn't built something that now needs to be "fixed". The team doing the fixing sees this as a problem, but the team who built the monstrosity got something built and into production and see it as a win.
It's entirely possible that a "proper" solution built by a dev team would have been superior. But that ultimately doesn't matter if the only way a thing sees the light of day is through the Excel/Visual Workflow Tool pipeline.
These products are not targeted at people who have the ability to build things from scratch the "right" way. And the reasons people/companies buy them usually fall into a few categories: resource constraints, politics, and managers/directors wanting to gain autonomy to build things for their teams without fighting for budget/prioritization with central dev/IT.