Hacker News Reader

Comment by teddyh | Hacker News Reader

“While Excel 5 was being designed, Lotus had shipped a “new paradigm” spreadsheet called Improv. According to the press releases, Improv was a whole new generation of spreadsheet, which was going to blow away everything that existed before it. […] Of course, Improv is now a footnote in history. […] Why? Because in Improv, it was almost impossible to just make lists. The Improv designers thought that people were using spreadsheets to create complicated multi-dimensional financial models. Turns out, if they asked people, they would discover that making lists was so much more common than multi-dimensional financial models, and in Improv, making lists was a downright chore, if not impossible.”

— Joel Spolsky

https://www.joelonsoftware.com/2000/05/09/the-process-of-des...

Archive.org link: https://web.archive.org/web/20170606181942/https://www.joelo...

View on HN

Disclosure: Spolsky worked on Excel for several years as a product manager, and is biased.

The other reason that Microsoft Office tools succeeded is that Microsoft was very good at all of the extra-technical stuff – business deals, marketing, software bundling, legal fights, etc., some of it quite underhanded/anticompetitive.

Lotus made Improv for NeXTSTEP originally, so there was a niche platform problem for the first 2 years, and after it had been ported to Windows Improv was competing against Lotus’s own previous spreadsheet program 1-2-3, creating consumer confusion and confusion within other parts of the Lotus organization.

But the main reason is probably just inertia. Once people have learned one tool that seems to work well enough for their needs, getting them to adopt something else (especially if it isn’t conceptually identical) is twice as hard.

He probably is biased - but it seems they were building "spreadsheet for advanced users" when most people who use a spreadsheet are not or will ever be advanced users.

Yet there is an enormous amount of important money-work being done by Excel users who are far more advanced than me, building what are essentially computer programs / databases.

So even if that market has fewer users then Excel itself, there an still be money to be made from people who could use a tool that is better behaved than a consumer spreadsheet but easier to use (for domain specialists) than a general programming environment.

This is the argument that (IMO, rightly) people make against the continued use of C for systems programming.

But C aficionados insist they understand its behaviors well enough to not get bitten...

I accept the analogy here between Excel and C, but I think the moral is the opposite.

Excel is very likely a good choice for a lot of people doing complicated stuff with it -- in spite of its problems. It's just that some of those people might also want a niche product that replaces it for some particular purpose. Not an entirely new, general purpose tool.

The problem with using a tool with say 5% of the features excel provides is you sooner or later run into some feature you need, but is absent in the niche tool. You can't exactly split your work between two tools now. So it just makes sense to use the tool with the super set of all the features, even if you don't use all those features all the time.

The right comparison for this would be using some thing like awk/sed compared to using Perl.

Apart from “home” use, Excel users usually lack functionality and tend to build an unmaintainable mess in their spreadsheets. Quarter of my ‘system integration’ work came from Excel users trying to automate and/or formalize what they created in there for years. Sadly, these types of businesses take hell of efforts to convert to good information systems; high failure rates are normal there. For me, Excel is a heroin of a business world. It drags you from reality and breaks you when you’re trying to get back into it.

>>Excel users usually lack functionality and tend to build an unmaintainable mess in their spreadsheets

Excel needs to be viewed as a programming tool. And a good programming tool is something that gives as much control to the user.

There are many Excel tutorials online and on Youtube that try to teach you how you can do MVC with Excel. That is separate presentation from data, and its control.

There aren’t enough trained programmers in the world to do in code what all Excel users get done with spreadsheets. Let alone programmers who understand the business of the company they work for.

It's the dream Microsoft Access should've been, but for whatever reason never caught on.

Microsoft tried it a few times, the latest attempt was LightSwitch, recently killed in favor of PowerApps.

https://blogs.msdn.microsoft.com/jasonz/2010/08/03/introduci...

https://powerapps.microsoft.com/en-us/

The thing is, with Excel the programming is gradual, you first just add some formulas, followed by conditional formatting, small macros, until eventually jumping into full VBA.

Alternatives, including Access, require good programming knowledge from day one.

But I still fundamentally agree with this approach. Perhaps not all users, but most users should be given a path toward more programming based approach to modelling.

I think lighswitch was stillborn when it wasn't distributed as part of office (irrespective of its merits, haven't tried it myself). Asking IT departments to deploy it was bound to fail. It needs to be available at the fingertips of excel users, like VBA is, always there, running with Excel, and you know it is available to everyone else so anything you do will run without having to install anything on their machine.

Powerapps look amazing to be honest. I actually would like to use that and the BI integration, but unfortunately I don't have enough reason to spend money on it. The small organising that I do need ends up in Airtable instead, but that's not even close. (I miss proper graphing features)

Spreadsheets exist because it would take programmers weeks to produce, what it would take a afternoon to do it in excel.

Plus most of the time these are one time throwaway efforts done to quickly arrive at results and make decisions.

It will all begin to make sense when we think of Spreadsheet work as programming and not something separate from it.

Oh god, this 'is biased' trope again. It's tiring. You can just say the first clause of that sentence and the meaning is the same.

I used to work in support for Lotus in the UK. A few people used Improv, but if IIRC once you got beyond a few dimensions the file sizes became huge.

The other alleged reason for Office beating our SmartSuite is the way new Windows or MS Dos releases always seemed to accidentally break Ami Pro or 1-2-3.

Maybe improv was just ahead of it’s time. It wouldn’t be the first time we see a product fail only to have a similar one years later become a success. I bet there are more people using excel for complex stuff now, they might as well find a niche.

I think Improv was only solving one of the use-cases that you'd use a spreadsheet for - and thats why it was never widely adopted.

Yes. Kind of like saying that you could replace a general programming language with FoxPro because all you ever need to do is database related code. As a programmer, even if you only deal with database, you kind of like to have full flexibility to do anything you want and not be cornered in a specific use case and fight against the product if you want to do things slightly differently. Users think the same way, even if it means more opportunities to shoot oneself in the foot.

This is spot on.

The biggest reason why lists matter so much is because. Most data structure use cases in everyday life are lists, or lists of lists(Tables). Of course you have graphs and trees. But for all practical purposes, most problems there are, are either lists or tables. Excel will benefit, if they add optional tree and graph interfaces. Like you get an option to build a graph of how your business things are connected and help you figure the shortest steps to a solution etc.

And yes if you want to build a Microsoft Excel killer. Please build a thick client desktop application which is as feature rich as a Excel and give it away for free. Charge for the cloud usage. Anything less isn't going to cut it.

Excel is just way ahead of anything that is out there. The only competing product I've seen is Libreoffice Calc. But even that has a long way to go.

Also, you live in a 3D world, hence any non-cognitively overloading "work surface" has to be -1 dim, so 2D, hence List or ListOfLists (spreadsheet). Anything else will require brainpower you could be using for something else instead! (like solving the actual problem, not masturbating with the tools...)

Other alternatives are Trees, which are also "lists of lists" but without the size constraints (Graphs "look 2D" but easily grow multi-dimensional information-wise). But trees have a big problem: they easily grow bigger than the screen, so you expand/collapse them, so you always end up with hidden information that you can overlook - hence they are really bad for sloppy people (I know it from first hand experience... if I'm not careful I can really forget that whole sections or Wrkflowy trees or Org-mode documents I work with even exits!)

Now combining list & lists-of-lists & trees & formatted text could work, if they manage to not informationally overload the user.

But I'm 99% sure that any tool more complex than matrixes and trees as information shape will not take on wide scale, until we manage to couple it with some superhuman-Clippy AI-assistant that almost reads your mind. We're stupid monkeys and it's only so much our brains can handle...

Slightly off-topic, but speaking of lists...

http://www.spiegel.de/international/zeitgeist/spiegel-interv...

Demo of Improv on NeXT:

http://www.youtube.com/watch?v=TbsfvdZXE7s

Seems a lot like PivotTables crossed with a normal spreadsheet.

That's actually pretty cool!

Pivot Tables were invented by Lotus: https://en.wikipedia.org/wiki/Pito_Salas

Thanks. It’s interesting but even that solves a very specific use of Excel. It will become painful if you need to cross reference different periods, etc. I sort of built a similar calculation engine for work that works in a similar way, but it is a solution taylored to a very specific problem. Finance depart who would be the primary user for that would still need Excel on the side.

Reply on news.ycombinator.com