Hacker News Reader

Comment by dang | Hacker News Reader

[stub for offtopicness]

Why are people getting hung up on his hubris? If his megalomania gives us Mathematica who cares. It's a phenomenal accomplishment. It can, out of the box, do hundreds of things that'd take you weeks in python/julia/etc or would be entirely impossible for most in other systems.

> Why are people getting hung up on his hubris? If his megalomania gives us Mathematica who cares.

We've had some bad results taking that approach recently - which we really should have anticipated, with millenia-old stories warning us about it. For one thing, megalomaniacs tend to be frauds (perhaps because, to be a megalomaniac, it's necessary to lie to yourself); maybe Mathematica will turn out to be the FTX, or Tesla auto-pilot, or ..., of mathematical software/languages.

No way - how could that happen? All those people at Mathematica would just go along with it, right? There's no evidence! It would be too brazen!

Mathematica has been publicly available for 35 years now, this is the 14th release. The tool has worked extremely well and been tangible for decades, with the value during that period being in the customer's direct use. In what ways does this remind you of FTX or Tesla auto-pilot?

Well, he claimed that Mathematica, particularly cellular automata implemented in Mathematica, would bring about "A New Kind of Science" as described in his thick book of that name. It didn't of course -- people like to still play with cellular automata like The Game of Life (or my favorite, Wireworld), but no revolution in science involving CAs has occurred. The whole thing was a marketing ploy for Mathematica disguised as science. Not mention that the only truly new thing mentioned in the book (that CAs can be Turing complete) wasn't even Wolfram's finding, but rather Matthew Cook's (who sued him for not crediting it to him, although they have since settled).

Wolfram does lots of wacky stuff but how does it make Mathematica any less tangibly useful today? He can want to use Mathematica as part of eradicating cancer using toilet brushes and it would still not change that Mathematica is and has been an extremely useful piece of software for decades. The point isn't every single thought he has is gospel revolution it's that Mathematica is already a delivered value.

That's still not very adequate comparison.

Musk has been promising self driving for... what 10 years now?

Hyping a working tool like Mathematica that will bring a revolution is a vague hope rather than a concrete promise of outcome.

Musk doesn't do this for you.

Notice the pattern, he does it to attract the kind of engineers attracted to solving 'unsolvable' problems.

Whether that's a good strategy is up in the air.

ah yes mastermind musk defense.

He is not 'lying' he is 'envisioning future to motivate engineers'.

He’s totally lying, just not to you.

Good engineers in specialized fields are really difficult to hire, this might not be apparent to people in the software industry.

> megalomaniacs tend to be frauds

Eeeh. Scott Aaronson review [0] and Cosma Shali's review [1] of NKS kinds of points to this direction, so he may well be a fraud. The plagiarism case regarding rule 110 (and attempts to hide this through NDA and lawsuits [2]) doesn't do him any favors, either. Indeed maybe the biggest problem of Wolfram isn't the megalomany per se, but the total unwillingness to give credit to others and cite their damn papers. Wolfram simply isn't in the business of sharing his bibliography, which is problematic for a scientist.

However, the newer Wolfram Physics [3] [4] looks so damn promising that I'm willing to entertain possible quackery. I mean it surely has many ideas regarding how a digital universe would look like; he may be all wrong in the details but his ideas look important contributions to me. I sometimes think what's like to be at the frontier of science; today we take relativity (for example) for granted but there was a time when it was up in the air. When I read Wolfram's stuff stuff I just think this could very well be true, and while there's absolutely no evidence for it the conjectures all make sense.

[0] https://arxiv.org/abs/quant-ph/0206089

[1] http://bactra.org/reviews/wolfram/

[2] https://news.ycombinator.com/item?id=13961947

[3] https://www.wolframphysics.org/

[4] https://writings.stephenwolfram.com/2020/04/finally-we-may-h...

Thanks for introducing some knowledge and fact to the discussion.

> However, the newer Wolfram Physics [3] [4] looks so damn promising that I'm willing to entertain possible quackery.

I don't know him, and clearly you know quite a bit. Still, that tradeoff is exactly where we've made our mistakes - the temptation of that payoff.

> maybe Mathematica will turn out to be the FTX, or Tesla auto-pilot, or ..., of mathematical software/languages.

You can install Mathematica at any time. It produces graphical output. You can see for yourself what it does, and what it does not do.

If Wolfram were making claims about a future unreleased version of Mathematica, sure, I would absolutely weigh his ego against those claims. But they are largely irrelevant when it comes to a currently existing and available product. If he were to claim that Mathematica 14 cures cancer, that would deserve eyerolls but it wouldn't tarnish the quality of the software.

> We've had some bad results taking that approach recently

I'm struggling to believe that Wolfram's hubris is anything like that of Trump or Musk or SBF etc. Wolfram seems like a genuinely smart person who knows he's smart, but is probably too aware of it. But he's not done anything majorly bad - and arguably has done good in creating tools for scientists and engineers. I think we need to give him a break. I certainly think that the perennial discussion of his ego is tiresome and distracting.

I heard this before, but never asked, what is so special about Mathematica that cannot be done by other softwares?

I'm basing this on my experience a few years back.

It's not that you can't do it with other software, Mathematica just has all of it included by default, with very comprehensive documentation. And a slick UI.

Sure you can probably do most of it in python, but you'll find yourself chasing some obscure modules to get some things working. Even just to get arbitrary precision calculations for just about everything, for example. And don't forget the importance of a documentation that explains every option with examples (barring some obscure stuff, which can be quite annoying if you encounter it).

Since it's all integrated Mathematica allows you to go from calculating Sin[2] to arbitrary precision, to calculating the derivative of Sin[2x], to showing the first 10 terms of its Taylor series. All using the same sine function.

This does have some downsides. For one it's a pretty heavy program to run. And because they want to include everything they need to be quite opinionated about certain things. There are multiple ways to define fractional derivatives for instance, since they include a function for it they must have picked one of them. And then there are the name clashes, I once had to laugh quite loudly when I tried "Rotate[{0,1}, 45deg]" and got back an image of {0,1} at a 45 degree angle.

It is an impressive piece of engineering. Which could have had much more of an impact if it was a bit more open, but oh well...

You should give it a spin. It’s an incredibly comprehensive Computer Algebra System with all the batteries included…. No, with a small nuclear reactor in the box.

I’m just sad that I don’t have access anymore after college, being too cheap to pay for a license.

It comes free with a Raspberry Pi.

https://www.wolfram.com/raspberry-pi/

For trying and basic work can also use a free Wolfram Cloud account.

Get then free kernel and WLJS notebook

- Far better language design, actually designed from ground up to write Mathematics in, rather than hacked in later.

- Specially, the language supports symbolic computation natively. In a lot of software, you have to declare symbols as symbolic variables before using them. In Mathematica, you don't have to deal with this nonsense.

- A library of both symbolic and numerical algorithms that is far far far better than any other library out there. Especially its outstanding how much symbolic computation algorithms are built in. Fairly fast numerical algorithms as well. Best of all, Mathematica guesses the best algorithm to use in a particular situation with frightening accuracy.

- A lot of Maths is just built in. Example: A number of common Groups are just there for you to immediately start playing with.

I had thought of him as megalomaniac, but I recently heard him do a two hour podcast on the Joe Walker Podcast (formerly Jolly Swagman). He came off as relatively normal for someone who earned a Ph.D. in particle physics from Caltech at the ripe age of 20.

> It’s only recently that I’ve begun to properly internalize just how broad the implications of having a computational language really are—even though, ironically, I’ve spent much of my life engaged precisely in the consuming task of building the world’s only large-scale computational language.

Exemplar of humility.

Wolfram should get a dedicated title tag (Wolfram) and hn rules should be amended with ‘this dude is a bit full of himself, everybody knows that, no need to bring it up’

Saying that as someone who’s consistently surprised by just how big his ego seems to be, getting bigger each year, didn’t think it’s possible.

This could be covered if there was a principle to respect the man in the arena. That should also discourage you-reinvented-the-wheel pile-ons.

The Ratatouille critic put this well - "We risk very little, yet enjoy a position over those who offer up their work and their selves to our judgement. We thrive on negative criticism, which is fun to write and to read."

When you've done as much as him any kind of ego that you may possess becomes justified.

The admin has already decreed exactly that. People just love to whine.

Or, for that matter, just come together in community and crack jokes we can all agree on, at the expense of someone we all know. Call it A New Kind of Socializing.

It sounds like once you get the zen of Mathematica, it’s a truly exceptional tool to work with.

It’s a shame that the only person it seems has get the zen of Mathematica is Stephen Wolfram himself.

In his case the zen of mathematica is that people will write a slick function for whatever he is interested in at the moment.

Everyone else needs to deal with stuff like '#1 + #2 & @@@ # &/ @'.

I mean, it really is good software. It's by far the best CAS around and the "standard library" is extremely impressive.

But like all software it involves tradeoffs. Fanatics of any technology or idea are never the easiest people to reason with.

Agreed, but what are the tradeoffs for Mathematica?

Mathematica has the best CAD capabilities for many/most tasks, but the language is rather horrible. E.g. code can not be commented, the program state is difficult to manage or even reset, it's very tied to the Mathematica "notebook". And it's quite idiosyncratic.

I sometimes have to use it where Sympy fails, and every time it's almost constant WTF.

And it's very proprietary and very expensive.

> code can not be commented

Mathematica has a C-like syntax for comments

    (* this is a comment *)

but it doesn't have C++ style comments

    // This is a C++ comment

Actually those are Pascal-like comments. Pascal uses (* ... *) for arbitrary comments as well as { ... } for single line comments.

I stand corrected. For some reason I hadn't figured this out earlier even though I tried to find it.

Note it's free on the raspberry pi!

https://www.wolfram.com/raspberry-pi/

If you have one collecting dust in a drawer, I'd heartily recommend checking it out.

check the license, though.

https://www.wolfram.com/legal/agreements/wolfram-mathematica...

> And it's very proprietary and very expensive

get free kernel and wljs notebook

Comes as a big monolithical block that's much harder to work with if you have to develop bigger systems, language tooling is lacking even compared to something like Python that's already mediocre, in many aspects (data analysis/data engineering, machine learning, etc.) isn't as good as other options like Python or R.

It's also not great as a general purpose language. It was never meant as such of course, but working in a real language where I can take stuff off a notebook and have it become a script has its advantages.

Finally, I'm willing to use open-source software that's slightly worse if the difference isn't huge, and for many use cases the difference isn't huge. For some it is, and that's where Mathematica is great.

It is a bit Zen, but if you don't use it for a few months you forget everything because the syntax is so different.

> In the arc of intellectual history it defines a broad, new, computational paradigm for formalizing the world.

I am in awe of what Wolfram achieved with Mathematica. And also the size of their ego.

So what if he likes to put things that way, there is hardly any competition to Mathematica, in what it is capable delivering.

Yet many of those that complain about Wolfram, will idolatrate Steve Jobs, and they aren't (weren't) that different in praising themselves.

I was honestly expecting Wolfram himself to have claimed the invention of LLMs or transformers, probably by saying it’s really a scaled up implementation of some other function in Mathematica.

Joking aside, he sort of implies this in his essay explaining GPT models[1]

    At some level this reminds one of the idea of universal computation (and my Principle of Computational Equivalence), but, as I’ll discuss later...

And his "Principle of Computational Equivalence" is [2]

    "There are various ways to state the Principle of Computational Equivalence, but probably the most general is just to say that almost all processes which are not obviously simple can be viewed as computations of equivalent sophistication."

Which a cynic might say is mighty convenient, because this is non-specific enough that you can apply it to basically anything and say you invented it and/or it's equivalent to something you invented and if it doesn't quite fit, you can use the "various ways to state" clause to weasel-word your way into something which does.

I find Stephen Wolfram frustrating for this reason. Benoit Mandelbrot is another guy who constantly seems to claim he invented everything, eg the Efficient Markets Hypothesis (which Mandelbrot basically claims he invented because he gave Eugene Fama some advice about the price process for stocks when Fama was a PhD student even though Fama/French was after that and Mandelbrot's idea of the price process for stocks is very obviously and demonstrably wrong if you know about market microstructure and/or look at trade marketdata[3]).

[1] https://writings.stephenwolfram.com/2023/02/what-is-chatgpt-...

[2] https://www.wolframscience.com/nks/p716--outline-of-the-prin...

[3] At a microstructure level price movements are made up of individual trades which jump around wildly with gaps in both the time and price dimension, and it has a base level beyond which you can't "zoom in" any more- it isn't some kind of fractal scaling in time and/or volatility which is what Mandelbrot wants it to be.

And the penny drops!

Thanks, I didn’t know he already did it. Totally on brand though.

Every mention of Stephen Wolfram deserves a link to Cosma Shalizi's epic take-down of a review ("A Rare Blend of Monster Raving Egomania and Utter Batshit Insanity"): http://bactra.org/reviews/wolfram/

What an amazing review-Thanks for that. It really brings together a lot of sad threads in my mind from that time. Having been the kind of guy who waited patiently for hours for home-made CAs to wiggle about on my VGA monitor and/or wrote little artificial life simulations etc for years in lonely isolation I bought "A New Kind of Science" with great expectation when it was in huge piles in every bookshop and was kind of devastated by how empty it was. I had really hoped for so much.

This is absurd and cannot be true.

Shmidhuber beat both of these guys easily with his paper from decades before these amateurs even claimed to start working on it.

And tehy didn't cite him. It's a travesty.

Are they wrong :) ?

It maybe be irrational, but the absolute arrogance of the author is off-putting and makes me reluctant to use any of his creations (which may be great I don't deny). I may have developed a heuristic that tells me to be very suspicious of people that are so sure and full of themselves as to become dangerous because they can be spectacularly wrong. It may have stemmed from the pandemic...

So, sure, you are smart and clever, but no.

Arrogance is not only annoying, but also indicative that the person isn't as good as they appear. Wolfram is different, however. Yeah, he's probably insecure due to some childhood-related reason, but don't let this trick you into believing he's just some random Internet person full of hot air. This guy is a genius and a fantastic explainer. His arrogance is just one of his well-documented personality quirks.

Just look the other way and try to learn something from him.

I recommend to watch some of his regular livestreams on Twitch [1], where I have learned that he is also a quite humble, honest and kind person. Albeit one who likes to hear himself talk (and is aware of that), but I also really enjoy his ramblings on science somehow.

With all the things he has accomplished and his competence across different fields, I don’t mind him being arrogant sometimes. He is working in the area of fundamentals of science, so of course there will be bold claims made - anyone can judge for themself if they hold water.

Science in institutions is often stuck in its entrenched paths and few people dare (or cannot afford) to step outside and try new things, especially across different fields. I believe people like Wolfram, who have earned the knowledge and experience to be able to make a serious attempt on advancing science in new and different ways, are a very valuable resource for humanity, even if they may not succeed.

[1]: https://livestreams.stephenwolfram.com

I don't care about arrogance, but closed source languages have no place in the 21th century IMO. I could use it if they open sourced at least the compiler

There's Mathics, a subset of the language implemented in Python, but unfortunately after the main author of that was hired by Wolfram, the project seems to have basically died. Still fun for what it is.

https://mathics.org/

Well, there is https://github.com/WolframResearch/codeparser

I'm happy that wolfram research decided to open source a few libraries, and it would be awesome if they eventually followed the steps of .NET Core and other formely closed ecosystems and opened up the whole toolchain

I never really got what made people so allergic to Wolfram. After the many outrages online I tried to dig and I never found anything shocking. There's not too many mention of "I" or too many self flattering hyperboles.. so I don't know.

> I never really got what made people so allergic to Wolfram. After the many outrages online I tried to dig and I never found anything shocking

For me it was trying to pass his employees work as if it were his own, through NDAs and lawsuits, like happened in this case:

https://en.m.wikipedia.org/wiki/Rule_110 https://www.complex-systems.com/abstracts/v15_i01_a01/

We only know this case because the employee fought back. Makes one wonder how many similar cases happened but never surfaced.

Oh interesting.

Previous discussion: https://news.ycombinator.com/item?id=13961947

He's actually very intelligent and has a record of producing a very successful and well loved product by many. He did things on his own terms and I can respect him for it, even if he would probably annoy the hell out of me in real life if we were just grabbing dinner.

Reply on news.ycombinator.com