Sunday assorted links

1. Jupyter vs. Mathematica, by Paul Romer.

2. The cost of being a “charismatic” animal.

3. IRS Says Fewer Than 100 People Have Reported Bitcoin Holdings So Far.

4. Should publishing move out of London to the north?

5. [India] declare[s] that a small amount of plagiarism—10% of a thesis, article, book, research paper, or other document—is acceptable, but that more extensive copying will result in increasingly severe punishments.” Link here.

6. Reputation inflation, or why the value of ratings tends to erode over time.

7. Do same-sex couples run a greater risk of poverty?


1) I have only started reading this piece, and it looks excellent so far. But I thought I would do a throwback to the old "open source is communism" days.

It really wasn't so long ago, okay maybe 20 years, but this really positive concept for innovation and change was opposed by people who couldn't fit it in their political idea map. If people were just sharing software, that was communism. All software should be sold by good Americans.

( I guess people who really remember those days will remember Microsoft assisting in the SCO lawsuit to end Linux once and forever.)

I guess it nets out to be a positive example, that change can happen in a relatively short time frame, and that a very strident reflexive conservatism can dissipate.

(Now Microsoft is a member of the Linux Foundation, and even open sources their own development tools.)

Times change.

Read the Atlantic link in the first paragraph as well.

The economics of proprietary software tend to make natural monopolies; maintenance and support costs are high, users are placing bets as to who will still be around in 5 years. That is why IBM and Microsoft had monopolies. Customers wanted it that way.

There are competing versions of node, competing javascript engines for browsers, competing development frameworks. The pace of development is quite remarkable.

Mathematica according to the Atlantic article has more polish and finish, which in theory would save users time and effort, but working around the limitations of the proprietary model swallows up any benefits.

I think there was a software guild aspect in the beginning. Some people were like "if you give away your software, how can we all make money?"

What people ultimately discovered was that is easy to make money selling software similar to free software, partly because of those benefits you describe, and partly because commercial software just has people out there selling.

So Microsoft ends up on the Linux Foundation without the fear that you all will switch to Linux on your desktop.

And maybe Wolfram will have a solid future selling proprietary advantage to Wall Street or something.

The Mathematica vs OO alternatives has the classic shape where Open Source has major advantages and ultimately wins: The core of the design is quite small, all things considered, and achievable with many languages, iPhyton included, and most of Mathematica's value added is precisely the kind of thing people do for free.

The economics of Open Source, are now very well understood: We are very far from The Cathedral And The Bazaar. There's the kind of products where OSS is a terrifying risk (Databases, software infrastructure and tooling) and the places where it's easy to make money selling software (Services, Big Data, Enterprise Tailored solutions). Every software-only startup has to answer why OSS won't eat them, and some companies have great answers: For example, payment companies are mostly software, but ultimately their business is, directly or indirectly, credit and insurance. Other times, we find companies existing in places where probably they shouldn't, and Wolfram is one of those. No matter how many beautiful demos Stephen gives in conferences, he doesn't have the economics of a platform, where users will give him improvements, or entire data sources, just because giving the work away will make their lives easier, while that is 100% true to Jupyter. The fact that he is the most arrogant person in the state of Illinois affects his recruiting efforts, but ultimately the nicest CEO would still have trouble in his situation: It's an 80s company trying to operate 30 years later, in ways that aren't all that different as they were back then, when his competition was quite feeble, and more open alternatives were free compilers for FORTRAN 77.

Money in analytics software is in very high touch sales, where the company has a lot of proprietary data and an analytics team that provides narrower tools that nobody would share: Think Cambridge Analytica, but with better tech, less fraud, and dedicated to things other than politics. There's plenty of startups doing this kind of thing out of Boston.

More economics in one comment than at the World Bank, it seems. What a terrible blog post, he's apparently never thought about commercial software, and is just deeply pissed off that the PDF export wasn't to his liking?

Python et. al. have completely eaten up what Matlab used to do for me, i.e. pushing numbers around. This makes sense for all the reasons you say. People with labs tell me that there's a lot of hardware integration for which you need Matlab, and its cost is a rounding error on top of the microscope etc... but that still seems just like a matter of time.

Mathematica seems less obvious. I mean Wolfram is deluded if he really thinks this is the one true language (although it's almost a lisp, and every generation of new languages is more lispy than the last) and I don't know where they make the bulk of their money. But for their original constituency of people who work in math departments, it still seems pretty hard to replace. For one example, it has by far the best library of special functions & their many properties, the competition is Abramowitz and Stegun. This doesn't strike me as the sort of thing geeks want to spend their weekends replacing.

I’m sort of troubled to see a top economist compare proprietary software developers to a ‘hoarde of vandals’ (who, BTW, were pretty good conquerors as these things go). He clearly resents an episode with Wolfram. I know little of this domain, but in image processing where Adobe (the ‘odious’ to some) dominates, there has been a vigorous attempt to displace the proprietary leader. The only efforts that get close are other proprietary softwares. Where you’re trying to buck innovation in a mature product where the questions, let alone the answers, are not obvious, capitalism and competition (‘vandals’) still work really well.

... Then R and R-Studio seem to be on track to displace a lot of the market for the proprietary Stats packages. Maybe it’s something about science and techne?

It really didn't help avoid the commie reputation when the face of Free Software was RMS.

The deStallmanization process (over about a decade, from presentation of CatB in 1997 by ESR, to Torvalds rejecting the idea of re-licensing Linux under the new GPL3 in 2007) made it easy to say, "No, the commies are over there, we're the pragmatic mainstream programmers."

5. Indicates all that is right and wrong about India.

Reveals the Indian penchant for hierarchy, ordering, and grading..

This pervades all aspects of Indian life.

Or, to connect it to the whole open source creativity story, maybe America too should have a 10% rule. Borrow no more than 10% of a copyrighted work and you aren't stealing, use less than 10% in your own work and you aren't plagiarizing. I would think that would unleash massive amounts of creativity and productivity.

In Brazil, one can reproduce 10% of a book ("doing a xerox") without legal problems.

This means that any scientific work coming from India can now be assumed to be plagiarized. The only possible response to this from the wider academic community I can see is to openly create a second tier for India.

Well 10% anyway. But from the consumer’s point of view, so what? If the information is reliable what difference does it make?

If they do not have the spine to draw a line at 0% what makes you think they will draw a line at 10%? This can only go one way. They will say, weeeelll, you know, 15% is pretty close to 10%. It wouldn't be fair to punish anyone so close to the line. And in fact 20% is not that far from 15%. And before you know it they will have done what they wanted to do all along which is forgive everyone.

The only clear line is zero.

From the consumer's point of view it will undermine the professions as it will penalize those of talent and original ideas as opposed to those who can just cut and paste. It will mean that wrong ideas will just get cited and cited and cited rather than challenged.

It is a grossly flawed system, but publications are still the dominant factor in hiring decisions in academia. And for this, it is not just accuracy, but originality, ability to perform scientific work, that matters. If we publish plagiarized work, we will see our universities fill with incompetents.

7. Intersectional grievance mongering. The U.S. government should kneecap them by refusing to collect such personal data like the French government refuses to do with respect to race.

6. What, you mean lesbian households feature a lotta drama and not a lotta money, kind of what you'd expect of, say, two single mothers?

There seems to be a cottage industry among social researchers of supplying empirical support to pithy remarks uttered by Mr. Sailer now and again.

Nothing wrong with social science studies to confirm that what we already think we know. And even better to publish the results, so the profession doesn't fall prey to novelty bias.

However, there was really no need to jump to "lotta drama, not a lot of money" and the smear against single moms when there's a simpler explanation that women tend to make less money than men, in general (and lesbians, in particular, face discrimination in housing, jobs, sentencing & arrest rates, etc).

and the smear against single moms

It wasn't a smear.

in general (and lesbians, in particular, face discrimination in housing, jobs, sentencing & arrest rates, etc).

In your imagination.

Are there statistics showing that lesbians have higher arrest rates? Or for that matter, higher rates of committing crimes?

Yes, higher than straight women. And gays lower than straight men.

Also for income statistics, any measure which shows a "wage gap" M>F has lesbians above the female average, and gays below the male one. Unsurprisingly, but good to check. And contrary to what subdee above claims.

There's significant discrimination in favor of women in the criminal justice system, so even if lesbians receive less favorable treatment because they are perceived to be more like men or do in fact behave more aggressively like men, then the abstract egalitarians should be happy with the result.

1. I dutifully read the article in The Atlantic last week (and mentioned it in a comment at this blog), but I had no idea what the article was actually about since I haven't a clue about Mr. Wolfram (never heard of him), Mathematica, or Jupyter; it was as if I was reading about some strange religious language that only believers understood. As I read the article I kept thinking to myself whether it's a good thing that scholars speak in a language only they understand. How much influence can they have if nobody else understands them? Ken Galbraith was influential in large part because of his command of the English language. The same could be said about Keynes. Alan Greenspan was influential in large part because he spoke gibberish that the media construed as wisdom. Trump's new economic spokesman, Larry Kudlow, has a common man's touch, but a twitch that screams snake oil salesman. Am I the only person who has noticed his twitch, a twitch I noticed the first time I saw him in the breakfast line at the Hyatt Regency in Atlanta over 30 years ago.

Lots of microaggressive, anti-disabled discourse here against those who twitch differently.

It's good to accept ignorance gracefully, it's not good to brag about it.

This is a poor comment. Jupyter and Mathematica and Python are just software tools, they aren't really "scholarly"; obviously researchers and scholar use them, but they are used by lots of non scholars too.

Python is a programming language, relatively recent (1991 -- thats mean it is probably younger than other programming languages you may have heard off : Basic, Pascal, C, Fortran etc.) and widely popular.

Mathematica is a symbolic calculator, propriety of Stephen Wolfram,
whose first version was put on the market in the eighties. Symbolic calculator means this : your normal calculator can compute 27.4 / 31.97 but you probably can't even ask it to compute (x+y)^2 - (x-y)^2. Mathematica will give you the correct answer 4xy. Of course, Mathematica does many more complicated things, in math, stats, science, etc.

Jupyter is an open-source recent alternative to Mathematica.

Jupyter is really a free alternative to the notebook front-end of Mathematica. Less polished, works in a web browser, but free. Similarly useful for exploratory calculations, saving what you typed and the graphs etc. produced all in one document.

The back-end was originally Python, now many other languages. These are great for some tasks, it's a huge ecosystem. Mathematica is also the back-end language invented to do symbolic calculations, and for this it's still extremely strong. The Python ecosystem is (IMHO) decades behind.

So true. I recently tried to do the stats of a whole paper using jupyter instead of R, and had to stop halfway when I ran into a dead-end with python’s available libraries. Good concept. Needs more development.

I've never used R, but I believe that's also both a language and a notebook front-end for it, right? Interesting that Python libraries don't do that well.

2. Why would polar bears be endangered? There's little human invasion of their natural territory. They eat seals, which don't seem to be endangered.

The typical American loves to see unusual animals on the Discovery Channel but not in the alley behind their house. Leopards. lions and tigers running loose in Phoenix or Knoxville would create a real emergency, like here. The relatively innocuous coyote is a big problem. Maybe because he keeps losing to the road runner he lacks charisma.

Trivia: I almost had to fight the coyote this week.

I was on my phone with my bank, staring out my kitchen window, when a neighbor lady went by with her small dog on a leash. Just a minute later a big coyote ran up and made a happy circle on my lawn before running after the lady. I stood there barefoot in my pajamas, thinking if she shouts I'm going to have to run out and fight the coyote! But she should didn't Shout and I guess the coyote thought better and went off a different way.

Given the population of coyotes and small dogs in the area, I have a walking stick by the door for this purpose. Not needed this far.

Get one of those walking sticks that hill-walkers used to use: they have a handy sharp spike on the end.

Don't bring a walking stick to a gun fight.

Lol, reminds me of a friend of mine who was involved in the construction of the Tehri Dam in 1994. The site was up in the mountains surrounded by forests. In the winters, when he stepped out to get to his car in the morning, he'd often run into tigers sunning themselves on his front porch. Obviously, that makes it difficult to leave for work on time. After a few encounters like this, he got a hose pipe attached by his front door and started hosing them down so they'd get out of the way. Turns out "charismatic animals" can be pests too!

Wow. I guess if a mountain lion shows up .. though that is a couple levels beyond!

6. This reminds me of what a very well-known criminal attorney in my city told me 40 years ago. Since I helped him secure his fees, what he told me had relevance to my work for him. He told me that when his client is acquitted, the client will announce to everyone that his attorney is the best attorney in America. Several days later the client will revise his assessment that his attorney did a pretty good job, but the client was innocent so the attorney's job wasn't all that difficult. Within a week, the client will revise his assessment that his attorney was lame, that if the attorney was any good the client would never have been charged in the first place. Thus, the need to secure fees up front. Of course, if the client is found guilty, well, even more need to secure the fees up front.

Fewer than 100 OUT OF THE LAST 250,000 returns that they looked at have reported Bitcoin transactions.

There's no duty to report Bitcoin holdings. There's a duty to report sales on Schedule D.

For now exchanges don't have a 1099-B requirement, but the IRS is requiring 1099-K reporting (that was the dispute between the IRS and Coinbase).

When researchers asked survey participants whether they thought these animals were endangered, explaining that they were not using IUCN terminology, they were surprised by the results.

So it turns out that when you tell poll subjects that words don't mean what they actually mean, they tend to act like words do mean what they actually mean anyways. "So why place polar bears on the list of "threatened" species if their numbers have been growing? Many scientists believe that due to climate change and resulting environmental factors, the trend is reversing. The Department of the Interior’s classification has more to do with increasing concerns for the future than with current population numbers."

Seems like a pretty subjective designation, at least for polar bears.

I expect it’s easier to get funds to study “threatened” species, and to punish the results, so there is an incentive to so label them.

“3. IRS Says Fewer Than 100 People Have Reported Bitcoin Holdings So Far.”

I find this almost utterly unbelievable, and suspect it’s just a new page in the old filing-season-scare-tactics playbook. There is no shortage of... let’s say “experienced“, rather than “sophisticated“... investors in the crypto currency space. Tax liability is a recurring discussion, in great detail, on dozens of blogs, with thousands of posts. I am aware of at least two entire businesses whose core model is assisting with tax accounting specific to crypto.

The only remotely plausible theory is that 99.99 percent will file at the deadline or later — which is a different non-story.

Yeah it is complete bullshit, they got caught lying earlier too.

"For the 2015 tax year, the IRS indicated that only 802 people had included cryptocurrency gains or losses in their tax filings."

That was also bullshit, a bunch of tax attorneys came forward and said they alone handled more clients than that, you can find some old podcasts that interview Tax Attorneys this question, it usually comes up and the shoot it down and bullshit, most likely they want more funding.

6. Tesla ratings...

7. in the years after the Great Depression the NYT ran a lot of articles on DINKs (double income no kids). Apparently, the wealth of DINKs was just a nice narrative.

#7 a. There's nothing more able to get a young person's sensibilities in order than when they are planning to start a family. I worked for a firm that preferred married men to single men because they were dependable - they had a family to feed.

b. Maybe there is some kind of stability in a mixed sex relationship. Each side evens out the extremes of the other.

The gender pay gap is actually a marriage pay gap where married men go out and earn a lot more money. They give up doing the things they like in order to work themselves into an early grave in return for .... well I am not sure. It used to be sex, respect and honor in the community. These do not apply any more.

So it is not a surprise that lesbians and Gay men do not do this. Why should they? It is not in the interests of married men either. Black men have largely given up on it too. It is a much smarter idea to do the things you love in exchange for less money - although society as a whole suffers when they do this.

Human beings are not particularly rational. It seems Gay men are more rational than heterosexual married men.

Not so long ago, men introduced their sisters to their single friends with good jobs. Not anymore. Since the government subsidizes sport sex and single motherhood, a guy with a good job, who could provide for a family, has no particular advantage over anybody else.

SMFS: poor men do barbecues, average men love cars, richer men look into boats, and even richer men......I have no firsthand knowledge of their tastes.

For sex, I've seem some relatives go for anything that looks remotely female. It's easy to get jealous when you only look at the most handsome romantic conquests. It's easy to feel sorry for them when you consider 1/20 is one woman you'll turn your head for.

The funny part is that values are not the most important issue. A poor person helping others will only be "selfless". To be "respectable and honorable", you have to help others and be damned wealthy.

#1: free and open source is the standard and expectation for this kind of software tool. That's old news. There are a variety of monetization strategies aside from simply selling software licenses. It would be nice if Mathematica moved to a FOSS model, but that's their choice to do or not to do. Wolfram Alpha offers a free to use tier. I use Wolfram Alpha for many things that are hard to do in other tools.

7. Gays have big medical bills, are sick a lot. Herpes, syphilis, gonorrhea, hepatitis... haven't even got to AIDS and HIV.

3) I call BS. I personally know two people who reported Bitcoin gains, and I'm not even marginally connected to the Bitcoin community. I suspect the IRS is not accurately counting the returns. There's no standardized asset identifier -- i.e. you could report "Bitcoin," "BTC," "Crypto," "Misc.," etc.

No way only 100 people have reported.

I thought the same thing, a closer reading of the article shows "Credit Karma Tax, ... that fewer than 100 of the 250,000 most recent tax filers through the service have reported cryptocurrency transactions." (i.e. Only 100 of 250,000 most recent Karam Tax users)

What does the Venn Diagram of crypto users and Karam Tax users look like comparability to the larger population? (Hint: probably not representative.)

( I reported also my gains also.. i don't remember what I put in as asset identifier)
but I am bitter, as I'm guessing the majority didn't report

4. seems weirdly redundant. People complaining about barriers to entry and costs of train tickets, when almost anyone can self-publish on Kindle.

1) So here is my conjecture about the question the article poses. Mathematica failed, despite technical accomplishments, because the norms of its developers clashed so obviously with the norms of its intended users. Jupyter is succeeding because the norms of the community that is developing it are aligned with the norms of its users.

Oh, for heaven's sake. Mathematica hasn't failed, and Jupyter isn't a direct competitor. And neither is a publishing format.

Mathematica is a functional programming language which allows some complicated and tedious mathematical operations to be done more easily. It's the industry leader, and has been basically since it was introduced several decades ago. It's not going away any time soon.

Jupyter is a frontend to Python (and now other languages). Python is a popular procedural programming language. There is one package for doing Mathematica style computer algebra with Python, but it's very rough. If you're using it for your job, a Mathematica license is well worth the money, despite the free competitor, and most universities and labs will have a site license.

Neither one is well suited for publication, because publication is about showing the results, not painstakingly stepping through each step of the argument. There's a reason that people don't publish every little subroutine of a code they used to perform a major calculation -- it's much too low-level to be of interest to the average reader.

Comments for this post are closed