I remain convinced that the reason most MS Office replacements struggle (beyond various anti-competitive behaviours) is that MS Office is really bad at what it does.
-
I remain convinced that the reason most MS Office replacements struggle (beyond various anti-competitive behaviours) is that MS Office is really bad at what it does. Word can’t do line breaking properly, makes semantic markup hard, doesn’t keep cross-references in sync automatically, and renders inconsistently between local and online versions. Excel interleaves functions and data in the most error-prone way possible and randomly interprets data as other types. PowerPoint is worse for producing animations than a 1996 edition of Flash.
None of these tools are good solutions for the problems that people apply them to.
You cannot displace a bad tool with a copy of the bad tool. People have little incentive to switch (until people start chanting ‘sovereignty’ as if it’s a magic spell) because you don’t solve their problems. People who are using the existing tool have a load of work arounds to deal with the fact that the tool is not a good fit for the tasks and any deviation in your behaviour from their current tool will break some of these.
You need to offer something that is actually a good solution to their problems. This is what I would build:
First, a good vector drawing tool. Take a look at OmniGraffle for what this looks like (maybe a couple of releases ago. I haven’t used the latest one and Omni Group keeps making all of their products worse with each release). This is important because it is a building block for two other key parts.
Next, a semantic text editor that tightly integrates with a desktop publishing program (they can be two views in the same program). Word is used in two ways, first to feed into publishing flows (where it’s a bad fit because it doesn’t do semantic markup well) and second as a DTP tool. But it’s a terrible DTP tool. Aldus Pagemaker on Windows 3.11 was better! The absolute minimum you want from a DTP tool is to be able to define to text boxes and have text flow between them. Word can’t even do that. But this is where you build on the vector editor: it defines page layouts and the text is flowed there. Ideally also provide an Access-like form tool, so people can fill in the top-level semantic markup without realising. You want to write a letter? Here’s a template that expects a to and from address, a salutation, a signature, and a message body. Fill in those bits and you get something properly typeset on company letterhead.
Add a Flash-style keyframe animation layer that uses the vector-drawing core. Now you have a better presentation tool than anything that exists today. Provide the ability to add notes and export key frames as pages in a PDF, and export to something web based for presenting.
The spreadsheet is the most interesting bit. You want something that behaves somewhere between Lotus Improv and Jupyter Notebooks: non-destructive data storage, generated columns, and a language that’s as easy to use as Excel’s calc language, ideally with the ability to plug in other languages for richer things. Rendered tables and graphs feed into the vector tool, where they can be used in the DTP model for print or the animation model for presentation. Provide a rich templating layer here, so companies can build the kind of data-processing that they want and have employees just fill in the data without touching (or being able to easily accidentally modify) the formulae, then have that feed directly into the rendering flow to produce reports.
Notice how little this looks like LibreOffice, OnlyOffice, and so on?
-
I remain convinced that the reason most MS Office replacements struggle (beyond various anti-competitive behaviours) is that MS Office is really bad at what it does. Word can’t do line breaking properly, makes semantic markup hard, doesn’t keep cross-references in sync automatically, and renders inconsistently between local and online versions. Excel interleaves functions and data in the most error-prone way possible and randomly interprets data as other types. PowerPoint is worse for producing animations than a 1996 edition of Flash.
None of these tools are good solutions for the problems that people apply them to.
You cannot displace a bad tool with a copy of the bad tool. People have little incentive to switch (until people start chanting ‘sovereignty’ as if it’s a magic spell) because you don’t solve their problems. People who are using the existing tool have a load of work arounds to deal with the fact that the tool is not a good fit for the tasks and any deviation in your behaviour from their current tool will break some of these.
You need to offer something that is actually a good solution to their problems. This is what I would build:
First, a good vector drawing tool. Take a look at OmniGraffle for what this looks like (maybe a couple of releases ago. I haven’t used the latest one and Omni Group keeps making all of their products worse with each release). This is important because it is a building block for two other key parts.
Next, a semantic text editor that tightly integrates with a desktop publishing program (they can be two views in the same program). Word is used in two ways, first to feed into publishing flows (where it’s a bad fit because it doesn’t do semantic markup well) and second as a DTP tool. But it’s a terrible DTP tool. Aldus Pagemaker on Windows 3.11 was better! The absolute minimum you want from a DTP tool is to be able to define to text boxes and have text flow between them. Word can’t even do that. But this is where you build on the vector editor: it defines page layouts and the text is flowed there. Ideally also provide an Access-like form tool, so people can fill in the top-level semantic markup without realising. You want to write a letter? Here’s a template that expects a to and from address, a salutation, a signature, and a message body. Fill in those bits and you get something properly typeset on company letterhead.
Add a Flash-style keyframe animation layer that uses the vector-drawing core. Now you have a better presentation tool than anything that exists today. Provide the ability to add notes and export key frames as pages in a PDF, and export to something web based for presenting.
The spreadsheet is the most interesting bit. You want something that behaves somewhere between Lotus Improv and Jupyter Notebooks: non-destructive data storage, generated columns, and a language that’s as easy to use as Excel’s calc language, ideally with the ability to plug in other languages for richer things. Rendered tables and graphs feed into the vector tool, where they can be used in the DTP model for print or the animation model for presentation. Provide a rich templating layer here, so companies can build the kind of data-processing that they want and have employees just fill in the data without touching (or being able to easily accidentally modify) the formulae, then have that feed directly into the rendering flow to produce reports.
Notice how little this looks like LibreOffice, OnlyOffice, and so on?
@david_chisnall But isn't Lotus Improv a good counter-example why this rarely works? (or FrameMaker, or Fireworks, or image-based languages etc.)
You might benefit from something substantially different, but people usually want what they're used to. What was the last really successful paradigm-breaking product?
-
@david_chisnall But isn't Lotus Improv a good counter-example why this rarely works? (or FrameMaker, or Fireworks, or image-based languages etc.)
You might benefit from something substantially different, but people usually want what they're used to. What was the last really successful paradigm-breaking product?
But isn't Lotus Improv a good counter-example why this rarely works?
Improv wasn't killed by third-party competition, it was killed by Lotus deciding that 1-2-3 would be easier to sell. Ironically, Improv was more popular with accountants, who were the reason that the flat sheet abstraction was introduced in VisiCalc: to make them feel at home.
FrameMaker was a pro tool. Adobe still sells it, but a FrameMaker license cost as much as (or more than?) a complete MS Office license.
WordPerfect was interesting. It held on for a long time, but mostly WordPerfect 5.1: later versions copied MS Word in ways that their loyal customers hated.
What was the last really successful paradigm-breaking product?
Google Docs moved people from a model of exchanging files to a model of exchanging links to documents. That was a big shift and far more important for interoperability than anything else because there's no expectation that someone you collaborate with in Google Docs has the same programs or a program that can read the same file format as you: they just need to be able to open a web page. I think that's probably done more lasting damage to MS Office's lock in than anything ODF-related and is a big paradigm shift.
The web and mobile phones that ran apps are both huge paradigm shifts that I've seen. So has been the shift from command-line to GUI tools for serious work (when I was learning to computer, GUI tools were seen as toys, lots of people refused to use a GUI spreadsheet because you'd waste far too much computational power rendering the display instead of computing the sheet).
The move from storing all of your files locally, to some on-premises file server, to some service that integrates a web-based viewer (and, often, editor) is a big shift for office documents.
A lot of things that used to be in-house Access or VB front ends are now web apps, often SaaS rather than in-house things.
A lot of people have moved from using spreadsheets to using something like Pandas in Python for data analytics, it just hasn't hit mainstream because Python is a big barrier to entry (less so than, say, C++, but much more than Excel's Calc language).
These changes typically take a decade to happen. Young people enter the workforce knowing how to do things a better way. They're forced to use the old way and so are familiar with both. When they reach decision-making levels in companies, they push for change and know both models well enough to be able to tell everyone why the new one is better.
Anyone who has serious experience with M365, Google Docs, and LibreOffice will tell you that they all suck in subtly different ways, they won't say that you absolutely must switch to one because it will reduce errors or improve efficiency in any important workflow (they may advocate switching to avoid lock, but that's a different dimension).
-
I remain convinced that the reason most MS Office replacements struggle (beyond various anti-competitive behaviours) is that MS Office is really bad at what it does. Word can’t do line breaking properly, makes semantic markup hard, doesn’t keep cross-references in sync automatically, and renders inconsistently between local and online versions. Excel interleaves functions and data in the most error-prone way possible and randomly interprets data as other types. PowerPoint is worse for producing animations than a 1996 edition of Flash.
None of these tools are good solutions for the problems that people apply them to.
You cannot displace a bad tool with a copy of the bad tool. People have little incentive to switch (until people start chanting ‘sovereignty’ as if it’s a magic spell) because you don’t solve their problems. People who are using the existing tool have a load of work arounds to deal with the fact that the tool is not a good fit for the tasks and any deviation in your behaviour from their current tool will break some of these.
You need to offer something that is actually a good solution to their problems. This is what I would build:
First, a good vector drawing tool. Take a look at OmniGraffle for what this looks like (maybe a couple of releases ago. I haven’t used the latest one and Omni Group keeps making all of their products worse with each release). This is important because it is a building block for two other key parts.
Next, a semantic text editor that tightly integrates with a desktop publishing program (they can be two views in the same program). Word is used in two ways, first to feed into publishing flows (where it’s a bad fit because it doesn’t do semantic markup well) and second as a DTP tool. But it’s a terrible DTP tool. Aldus Pagemaker on Windows 3.11 was better! The absolute minimum you want from a DTP tool is to be able to define to text boxes and have text flow between them. Word can’t even do that. But this is where you build on the vector editor: it defines page layouts and the text is flowed there. Ideally also provide an Access-like form tool, so people can fill in the top-level semantic markup without realising. You want to write a letter? Here’s a template that expects a to and from address, a salutation, a signature, and a message body. Fill in those bits and you get something properly typeset on company letterhead.
Add a Flash-style keyframe animation layer that uses the vector-drawing core. Now you have a better presentation tool than anything that exists today. Provide the ability to add notes and export key frames as pages in a PDF, and export to something web based for presenting.
The spreadsheet is the most interesting bit. You want something that behaves somewhere between Lotus Improv and Jupyter Notebooks: non-destructive data storage, generated columns, and a language that’s as easy to use as Excel’s calc language, ideally with the ability to plug in other languages for richer things. Rendered tables and graphs feed into the vector tool, where they can be used in the DTP model for print or the animation model for presentation. Provide a rich templating layer here, so companies can build the kind of data-processing that they want and have employees just fill in the data without touching (or being able to easily accidentally modify) the formulae, then have that feed directly into the rendering flow to produce reports.
Notice how little this looks like LibreOffice, OnlyOffice, and so on?
@david_chisnall totally agree. I think it is useful to remember that:
Word has an interface based on business letters.
PowerPoint has an interface based on overhead sheets.
Excel has an interface based on physical accounting spreadsheets.In other words, it's made for the office of 1994, where the printer and fax machine was king. It's time to leave that behind.
I think La Suite Numérique is quite interesting and seems to be heading in the right direction.
-
But isn't Lotus Improv a good counter-example why this rarely works?
Improv wasn't killed by third-party competition, it was killed by Lotus deciding that 1-2-3 would be easier to sell. Ironically, Improv was more popular with accountants, who were the reason that the flat sheet abstraction was introduced in VisiCalc: to make them feel at home.
FrameMaker was a pro tool. Adobe still sells it, but a FrameMaker license cost as much as (or more than?) a complete MS Office license.
WordPerfect was interesting. It held on for a long time, but mostly WordPerfect 5.1: later versions copied MS Word in ways that their loyal customers hated.
What was the last really successful paradigm-breaking product?
Google Docs moved people from a model of exchanging files to a model of exchanging links to documents. That was a big shift and far more important for interoperability than anything else because there's no expectation that someone you collaborate with in Google Docs has the same programs or a program that can read the same file format as you: they just need to be able to open a web page. I think that's probably done more lasting damage to MS Office's lock in than anything ODF-related and is a big paradigm shift.
The web and mobile phones that ran apps are both huge paradigm shifts that I've seen. So has been the shift from command-line to GUI tools for serious work (when I was learning to computer, GUI tools were seen as toys, lots of people refused to use a GUI spreadsheet because you'd waste far too much computational power rendering the display instead of computing the sheet).
The move from storing all of your files locally, to some on-premises file server, to some service that integrates a web-based viewer (and, often, editor) is a big shift for office documents.
A lot of things that used to be in-house Access or VB front ends are now web apps, often SaaS rather than in-house things.
A lot of people have moved from using spreadsheets to using something like Pandas in Python for data analytics, it just hasn't hit mainstream because Python is a big barrier to entry (less so than, say, C++, but much more than Excel's Calc language).
These changes typically take a decade to happen. Young people enter the workforce knowing how to do things a better way. They're forced to use the old way and so are familiar with both. When they reach decision-making levels in companies, they push for change and know both models well enough to be able to tell everyone why the new one is better.
Anyone who has serious experience with M365, Google Docs, and LibreOffice will tell you that they all suck in subtly different ways, they won't say that you absolutely must switch to one because it will reduce errors or improve efficiency in any important workflow (they may advocate switching to avoid lock, but that's a different dimension).
@david_chisnall Alright, moving to the network has been a big shift, despite the UI trying very hard to make it seem less so. The web truly was a big exception, as there was no good existing thing to map it to.
But on a more quotidian level, I'm not entirely sure that most of these shifts are on the same level as going from "just make it large and bold" to semantic text editing would be.
There often were quite a few intermediate steps there, and also quite often an attempt to copy the status quo. Google Docs as a collaboration is new-ish, but the editing interface itself even in the beginning was quite close to what you were used to when you open Word for Windows on your eMachine. It wasn't jottit or MediaWiki that won.
I'm very dubious about the success chances of anything that adds structure. It seems making anything more ad hoc tends to win out. Manual text formatting. Just placing numbers in cells. Placing text boxes on a sheet vs attaching meaning to any of it. Picking recently used documents or clicking on links in your email instead of navigating through folders. Or just tell your RNG-in-a-suit what you want to do.
(As for workforce shifts, I see more loss of information there than integration of various traditions/generations)
-
@david_chisnall Alright, moving to the network has been a big shift, despite the UI trying very hard to make it seem less so. The web truly was a big exception, as there was no good existing thing to map it to.
But on a more quotidian level, I'm not entirely sure that most of these shifts are on the same level as going from "just make it large and bold" to semantic text editing would be.
There often were quite a few intermediate steps there, and also quite often an attempt to copy the status quo. Google Docs as a collaboration is new-ish, but the editing interface itself even in the beginning was quite close to what you were used to when you open Word for Windows on your eMachine. It wasn't jottit or MediaWiki that won.
I'm very dubious about the success chances of anything that adds structure. It seems making anything more ad hoc tends to win out. Manual text formatting. Just placing numbers in cells. Placing text boxes on a sheet vs attaching meaning to any of it. Picking recently used documents or clicking on links in your email instead of navigating through folders. Or just tell your RNG-in-a-suit what you want to do.
(As for workforce shifts, I see more loss of information there than integration of various traditions/generations)
@mhd @david_chisnall
@visidata seems fairly popular (but not comparable to Excel, of course). It seems to have some Improv-like concepts (have not used Improv): original data are supposed to be immutable (although cell values can be edited), and you create filtered sheets and typed columns based on Python expressions, with some simplified syntax for column operations. -
I remain convinced that the reason most MS Office replacements struggle (beyond various anti-competitive behaviours) is that MS Office is really bad at what it does. Word can’t do line breaking properly, makes semantic markup hard, doesn’t keep cross-references in sync automatically, and renders inconsistently between local and online versions. Excel interleaves functions and data in the most error-prone way possible and randomly interprets data as other types. PowerPoint is worse for producing animations than a 1996 edition of Flash.
None of these tools are good solutions for the problems that people apply them to.
You cannot displace a bad tool with a copy of the bad tool. People have little incentive to switch (until people start chanting ‘sovereignty’ as if it’s a magic spell) because you don’t solve their problems. People who are using the existing tool have a load of work arounds to deal with the fact that the tool is not a good fit for the tasks and any deviation in your behaviour from their current tool will break some of these.
You need to offer something that is actually a good solution to their problems. This is what I would build:
First, a good vector drawing tool. Take a look at OmniGraffle for what this looks like (maybe a couple of releases ago. I haven’t used the latest one and Omni Group keeps making all of their products worse with each release). This is important because it is a building block for two other key parts.
Next, a semantic text editor that tightly integrates with a desktop publishing program (they can be two views in the same program). Word is used in two ways, first to feed into publishing flows (where it’s a bad fit because it doesn’t do semantic markup well) and second as a DTP tool. But it’s a terrible DTP tool. Aldus Pagemaker on Windows 3.11 was better! The absolute minimum you want from a DTP tool is to be able to define to text boxes and have text flow between them. Word can’t even do that. But this is where you build on the vector editor: it defines page layouts and the text is flowed there. Ideally also provide an Access-like form tool, so people can fill in the top-level semantic markup without realising. You want to write a letter? Here’s a template that expects a to and from address, a salutation, a signature, and a message body. Fill in those bits and you get something properly typeset on company letterhead.
Add a Flash-style keyframe animation layer that uses the vector-drawing core. Now you have a better presentation tool than anything that exists today. Provide the ability to add notes and export key frames as pages in a PDF, and export to something web based for presenting.
The spreadsheet is the most interesting bit. You want something that behaves somewhere between Lotus Improv and Jupyter Notebooks: non-destructive data storage, generated columns, and a language that’s as easy to use as Excel’s calc language, ideally with the ability to plug in other languages for richer things. Rendered tables and graphs feed into the vector tool, where they can be used in the DTP model for print or the animation model for presentation. Provide a rich templating layer here, so companies can build the kind of data-processing that they want and have employees just fill in the data without touching (or being able to easily accidentally modify) the formulae, then have that feed directly into the rendering flow to produce reports.
Notice how little this looks like LibreOffice, OnlyOffice, and so on?
@david_chisnall Regarding Excel, I've switched to grist: https://www.getgrist.com/
It's Excel but you use Python as the formatting language and it enforces a strict seperation between data and visualization. Also you can name your columns and all rows in a column have to use the same formula.
But the average excel user probably likes the flexibility of excel.
-
But isn't Lotus Improv a good counter-example why this rarely works?
Improv wasn't killed by third-party competition, it was killed by Lotus deciding that 1-2-3 would be easier to sell. Ironically, Improv was more popular with accountants, who were the reason that the flat sheet abstraction was introduced in VisiCalc: to make them feel at home.
FrameMaker was a pro tool. Adobe still sells it, but a FrameMaker license cost as much as (or more than?) a complete MS Office license.
WordPerfect was interesting. It held on for a long time, but mostly WordPerfect 5.1: later versions copied MS Word in ways that their loyal customers hated.
What was the last really successful paradigm-breaking product?
Google Docs moved people from a model of exchanging files to a model of exchanging links to documents. That was a big shift and far more important for interoperability than anything else because there's no expectation that someone you collaborate with in Google Docs has the same programs or a program that can read the same file format as you: they just need to be able to open a web page. I think that's probably done more lasting damage to MS Office's lock in than anything ODF-related and is a big paradigm shift.
The web and mobile phones that ran apps are both huge paradigm shifts that I've seen. So has been the shift from command-line to GUI tools for serious work (when I was learning to computer, GUI tools were seen as toys, lots of people refused to use a GUI spreadsheet because you'd waste far too much computational power rendering the display instead of computing the sheet).
The move from storing all of your files locally, to some on-premises file server, to some service that integrates a web-based viewer (and, often, editor) is a big shift for office documents.
A lot of things that used to be in-house Access or VB front ends are now web apps, often SaaS rather than in-house things.
A lot of people have moved from using spreadsheets to using something like Pandas in Python for data analytics, it just hasn't hit mainstream because Python is a big barrier to entry (less so than, say, C++, but much more than Excel's Calc language).
These changes typically take a decade to happen. Young people enter the workforce knowing how to do things a better way. They're forced to use the old way and so are familiar with both. When they reach decision-making levels in companies, they push for change and know both models well enough to be able to tell everyone why the new one is better.
Anyone who has serious experience with M365, Google Docs, and LibreOffice will tell you that they all suck in subtly different ways, they won't say that you absolutely must switch to one because it will reduce errors or improve efficiency in any important workflow (they may advocate switching to avoid lock, but that's a different dimension).
key bit you also referred to before: rendering differences
for decades that was an insurmountable bugbear -- until MS themselves shrugged it away w/ their own incompetence, when it became a complete non-issue
-
I remain convinced that the reason most MS Office replacements struggle (beyond various anti-competitive behaviours) is that MS Office is really bad at what it does. Word can’t do line breaking properly, makes semantic markup hard, doesn’t keep cross-references in sync automatically, and renders inconsistently between local and online versions. Excel interleaves functions and data in the most error-prone way possible and randomly interprets data as other types. PowerPoint is worse for producing animations than a 1996 edition of Flash.
None of these tools are good solutions for the problems that people apply them to.
You cannot displace a bad tool with a copy of the bad tool. People have little incentive to switch (until people start chanting ‘sovereignty’ as if it’s a magic spell) because you don’t solve their problems. People who are using the existing tool have a load of work arounds to deal with the fact that the tool is not a good fit for the tasks and any deviation in your behaviour from their current tool will break some of these.
You need to offer something that is actually a good solution to their problems. This is what I would build:
First, a good vector drawing tool. Take a look at OmniGraffle for what this looks like (maybe a couple of releases ago. I haven’t used the latest one and Omni Group keeps making all of their products worse with each release). This is important because it is a building block for two other key parts.
Next, a semantic text editor that tightly integrates with a desktop publishing program (they can be two views in the same program). Word is used in two ways, first to feed into publishing flows (where it’s a bad fit because it doesn’t do semantic markup well) and second as a DTP tool. But it’s a terrible DTP tool. Aldus Pagemaker on Windows 3.11 was better! The absolute minimum you want from a DTP tool is to be able to define to text boxes and have text flow between them. Word can’t even do that. But this is where you build on the vector editor: it defines page layouts and the text is flowed there. Ideally also provide an Access-like form tool, so people can fill in the top-level semantic markup without realising. You want to write a letter? Here’s a template that expects a to and from address, a salutation, a signature, and a message body. Fill in those bits and you get something properly typeset on company letterhead.
Add a Flash-style keyframe animation layer that uses the vector-drawing core. Now you have a better presentation tool than anything that exists today. Provide the ability to add notes and export key frames as pages in a PDF, and export to something web based for presenting.
The spreadsheet is the most interesting bit. You want something that behaves somewhere between Lotus Improv and Jupyter Notebooks: non-destructive data storage, generated columns, and a language that’s as easy to use as Excel’s calc language, ideally with the ability to plug in other languages for richer things. Rendered tables and graphs feed into the vector tool, where they can be used in the DTP model for print or the animation model for presentation. Provide a rich templating layer here, so companies can build the kind of data-processing that they want and have employees just fill in the data without touching (or being able to easily accidentally modify) the formulae, then have that feed directly into the rendering flow to produce reports.
Notice how little this looks like LibreOffice, OnlyOffice, and so on?
@david_chisnall ah, but excel is pretty good for making character sheets for table top role-playing games!
Checkmate, atheists!
... Wait...
-
@david_chisnall But isn't Lotus Improv a good counter-example why this rarely works? (or FrameMaker, or Fireworks, or image-based languages etc.)
You might benefit from something substantially different, but people usually want what they're used to. What was the last really successful paradigm-breaking product?
The iPhone. 2007.
