The funny thing about all this is that I think - despite all his bluster - that Wolfram is basically right about how uniquely powerful this stuff is in terms of making advanced symbolic programming with complex data accessible and visible, and in providing a rich-text notation to express computations. I think that Mathematica genuinely captures a piece of what it would mean to have a computer as a mind-amplifier, a tool rather than a mere appliance, one which is frictionless and accessible enough to be usable by everyday people. I think he's actually on to something (as opposed to merely being on something).
...The problem is that it's attached to the rest of the Wolfram language - a painfully awkward Lisp-ish thingy dependent on an expensive closed-source platform and a standard library that seems to contain everything you could ever possibly need to do ... but which is so intimidatingly enormous that you can't keep track of it, which turns every program into a trek through the documentation in case there's something there that already does what you need (and then trying to figure out how to plumb all these bits together.) You can do virtually anything with the Mathematica standard library ... which is good, because trying to actually write de novo code is despair-inducing.
And Steve doesn't seem to be able to recognize that Paragraph 1 is ultimately crippled for widespread utility by Paragraph 2.
I do hope that eventually someone manages to take the notational insights of the Wolfram language and apply them to some other platform. There's a vague vision in my head of something that combines aspects and insights of Wolfram/Mathematica, HyperCard, Jupyter, and Excel[1] to create a truly flexible and accessible end-user programming environment.
[1] people really underrate Excel as a programming environment, honestly. Yes, it's crippled and leads people to produce massive gross un-debuggable hellsheets ... but there's reasons (beyond just "it was the only usable software that could be run on office computers") that end users with a problem to solve keep turning to it despite those flaws. The combination of reactive programming, data-first visibility, no hidden state, decent approximations to structured programming by way of click-drag-and-copy-paste, and being able to reference variables and values without needing to name them has some kind of magic to it. There's quite a bit of interesting research on how to take something like Excel and turn it into a non-crippled programming environment - spreadsheet-defined functions (including recursion, lambdas, and higher-order functions natively in the spreadsheet environment, without having to drop into VBA!), dynamic arrays, an alternate computation-first textual view that exists simultaneously with the data-first spreadsheet view, first-class complex data structures ...
On [1], you would probably like Mesh Spreadsheet. Spreadsheet-defined functions (via IPC), lambdas, dynamic arrays, text pane for showing and editing sheet-as-code, editing tables as a first-class data structure, all portable and free and based on JS as the formula language. https://github.com/chrispsn/mesh
> but which is so intimidatingly enormous that you can't keep track of it, which turns every program into a trek through the documentation in case there's something there that already does what you need (and then trying to figure out how to plumb all these bits together.)
...The problem is that it's attached to the rest of the Wolfram language - a painfully awkward Lisp-ish thingy dependent on an expensive closed-source platform and a standard library that seems to contain everything you could ever possibly need to do ... but which is so intimidatingly enormous that you can't keep track of it, which turns every program into a trek through the documentation in case there's something there that already does what you need (and then trying to figure out how to plumb all these bits together.) You can do virtually anything with the Mathematica standard library ... which is good, because trying to actually write de novo code is despair-inducing.
And Steve doesn't seem to be able to recognize that Paragraph 1 is ultimately crippled for widespread utility by Paragraph 2.
I do hope that eventually someone manages to take the notational insights of the Wolfram language and apply them to some other platform. There's a vague vision in my head of something that combines aspects and insights of Wolfram/Mathematica, HyperCard, Jupyter, and Excel[1] to create a truly flexible and accessible end-user programming environment.
[1] people really underrate Excel as a programming environment, honestly. Yes, it's crippled and leads people to produce massive gross un-debuggable hellsheets ... but there's reasons (beyond just "it was the only usable software that could be run on office computers") that end users with a problem to solve keep turning to it despite those flaws. The combination of reactive programming, data-first visibility, no hidden state, decent approximations to structured programming by way of click-drag-and-copy-paste, and being able to reference variables and values without needing to name them has some kind of magic to it. There's quite a bit of interesting research on how to take something like Excel and turn it into a non-crippled programming environment - spreadsheet-defined functions (including recursion, lambdas, and higher-order functions natively in the spreadsheet environment, without having to drop into VBA!), dynamic arrays, an alternate computation-first textual view that exists simultaneously with the data-first spreadsheet view, first-class complex data structures ...