I don't care that it's not open source, it's a unique technology. Actual algorithms solving actual problems correctly. None of that fuzzy search that you'd get if you used Google. It's an amazing tool in its domain.
I absolutely don’t contend that it can’t do incredible things, and I think a huge swath of developers would be eager to take advantage of these capabilities were it under more permissive circumstances. I’m also happy for you that you have the latitude and willingness to build software in a programming language toolchain that is not under your control, both technologically and legally. I know at the companies I’ve worked for, including it as a part of the product would not fly with so many departments, most notably legal, exec leadership, and most SW developers.
The reason for its dismissal goes a bit deeper than the simple fact that it’s not open source. We pay for other software (like E&M solvers), why not this? Because its use is not compartmentalized. It’s a programming language with the world’s most expansive standard library. Just as is the case with—say—Python, folks using this have no end to what they would build, including company IP. The code would absolutely not be portable, and your tech stack would be vendor-tied. That’s a danger zone for any serious enterprise.
My problem with it being closed-source is that it doesn't form a stable platform for my work. If they decide to change their product in incompatible ways in the future, I'm sunk. I have no recourse, I cannot fix the problem if it goes deep enough, and there are no other implementations.
I learned this lesson (albeit from a distance) when Microsoft killed off Visual Basic 6 and replaced it with Visual Basic .Net.
I've got no stake in this, but I'll note that from the claims Wolfram has made, they've got a 30 year track record in which notebooks going back to 1.0 still run in the latest version.
I don’t think being forced to upgrade is as much of a concern as receiving incompatible upgrades, or not receiving anything at all. Remember, Wolfram is trying to reach industry here, not just individual software developers.
I agree. Plus, you can't peer into the layers underneath your software – if I'm writing code in Python or R, everything underneath me is there for inspection, if needed.
A closed source solution, where all you have is documentation (potentially flawed), including manually authored change information, just feels archaic at this point.
A lot of people use AWS though, right? It's about trust. Not trust in a "Amazon are good people I trust them" but more like "Amazon incentives are aligned with mine for this project, and I trust that using their platform wont f' me up down the line".
Companies should derive that trust factor from other factors based on their unique circumstances. If you are Amazon's competitor for instance it is probably not wise to use AWS for anything, but if you are a startup it could be very wise to take a bit of lock-in to avoid rolling your own infrastructure. If you are an established company maybe use AWS but in a way that you could move to another provider without too much pain, but balance that against some of the goodies you get by staying locked in.
I guess the same nuanced considerations have to be made to decide to use Wolfram Engine or not. There is no clear answer that suits all end users.
I've been reading as much as I can on RMS's site recently because I want to expand my mind on these sorts of topics. I really want to understand how "free" software can work at a practical level, where you say "here is a piece of software, it's $1000 from me, and you are free to give it to your friends, modify it, and resell it, even to people who would have paid me $1000, but now they wont need to" and how that ties in with making a sustainable business. I think it takes some luck and good imagination to make such a business work by selling some by-product such as consulting services or whatever. But for a lot of companies non-free is required in my opinion to exist as a business. Happy to have my mind changed.
AWS is more of a commodity, right? You use it for compute and storage and networking, but how much of what you write is intrinsically tied to AWS? With what we're talking about here, there's a language and a bunch of libraries where, if the company decides to change things, your code stops working, and there's no way to make your code work again without rewriting some arbitrary amount of that code. Maybe the company decides that a whole kind of activity is now bad, and you can't do it at all on their platform.
With AWS, you can pack up and go to a different cloud provider, and get the same Linux and the same gcc and the same TCP/IP stack. There's probably some other code you'd have to change, but the essential core is non-proprietary and nobody can take it away from you. Abstracting further, if Linux decided to go in a weird direction, you can get a POSIX-conformant Unix-like OS from other groups, like FreeBSD or Illumos. The Linux codebase can be forked as well, providing an even stronger insurance relative to what happens when some company's or business unit's incentives no longer align with yours.
And that part about aligning incentives is powerful. In the closed-source world, it's also fragile: One company being bought out can shift things tectonically for all of the software that company made. It becomes a treadmill, or a movie routine of constantly jumping from platform to platform as each of them dies off or is pulled out from under you as "corporate synergy" realigns strategic chakras feng shuis your old codebase into worthlessness.
> fahre dann mit dem Taxi zu einer Villa am Stadtrand. Ein bedienster steht am Eingang,
If you are "seriously" using AWS: A lot. Value of AWS is not that it offers compute nodes. But a full catalogue of further services and tooling APIs and migrating of that can be a big project.
It's not unique in that there literally exist other computer algebra systems (CAS), sure. But it is unique in two meaningful ways:
1. It is arguably the most sophisticated CAS available. Open source tooling has crept up in performance and completeness in recent years, but Mathematica still dwarfs every open source system in performance and feature availability. Competing proprietary systems like Maple are capable of beating Mathematica in certain specific domains (like PDEs), but that leads me to my second point.
2. Mathematica isn't just a CAS. It also supports sophisticated data analysis and ingestion, visualization, (some) machine learning, natural language processing, speech recognition, signal processing, climatology, meteorology, geography, financial analysis, and limited forms of convex optimization.
Some things that Wolfram Language does that maxima doesn't: machine learning, image processing, audio processing, time series analysis, user interface construction, build-in data, geo computation, natural language processing....
Maxima is just a computer algebra system- one small part of what Wolfram does.
Are Maxima's CAS algorithms comparable to Mathematica's? I have tried to used Sage and other systems in the past, but Mathematica has always been better at finding solutions for what I tried to do. But I have never used Maxima for real.
Computing symbolic derivatives and integrals, simplifying algebraic expressions, arithmetic with arbitrary length integers, solving simple ODE, computing limits and series, things like that.
Maxima is a fine piece of software, and I’m a huge proponent of its development, but doesn’t compete in the breadth or depth of Mathematica (or Maple) at all.
It certainly competes due to the fact of being free software, which is a strong requirement of my workflow. I can try Mathematica and Maple (and love them), but ultimately I have to end using a free software solution.
