Technology’s language bias

One of the many underlying tensions in Sri Lanka is that between people fluent in English and those who aren’t. After independence and Sinhala made the country’s official language, education was converted to “Swabasha”.

But the elite kept English. It is the medium in most private and international schools. University courses are mostly in English, as are the offices of big corporations. There is a huge wage premium in an English education.

All of which have resulted in an inevitable pushback.

The debate over English education has been a backdrop to my last couple of months, as I’ve been learning to code and scraping the web.

Just the other day I was scraping the Sri Lankan parliament website, and commenting to my grandpa how the HTML is all in English. But how can this be?

This is the central point of a great Wired article:

In theory, you can make a programming language out of any symbols. The computer doesn’t care. The computer is already running an invisible program (a compiler) to translate your IF orinto the 1s and 0s that it functions in, and it would function just as effectively if we used a potato emoji 🥔 to stand for IF and the obscure 15th century Cyrillic symbol multiocular O ꙮ to stand for. The fact that programming languages often resemble English words like body or if is a convenient accommodation for our puny human meatbrains, which are much better at remembering commands that look like words we already know.

But only some of us already know the words of these commands: those of us who speak English. The “initial promise of the web” was only ever a promise to its English-speaking users, whether native English-speaking or with access to the kind of elite education that produces fluent second-language English speakers in non-English-dominant areas.

There are a couple of multilingual programming languages, and programming languages based on other natural languages, but they’re nowhere near as supported or extended.

Without a large community of fellow travellers there aren’t big archives of questions and answers to query when you yourself have a problem, or huge repositories of packages and modules to extend your code.

I’ve recently been studying natural language processing, which is another domain with a huge bias towards English. The letters and symbols of many languages aren’t even supported by Unicode.

The Wired article ends with a hopeful message that we might eventually have “Swahili HTML” and “Russian HTML” in addition to “English HTML” (rather than HTML being synonymous with English). The author notes that people learn coding better in their native tongues, and that European writing was once synonymous with Latin before branching into many vernaculars.

But there needs to be a blend of these answers. There likely won’t be any one small natural language-based computer language that can compete with a large one for usability and flexibility. But that doesn’t mean localised computer languages can’t be powerful, useful or a great pathway.

Until then a lot of the power of modern technologies will remain the privilege of those with English language skills.

We’re really bad at reality

Have you ever tried to mock up the outcome from a series of coin flips? It turns out that we are horrible at simulating randomness.

But as a great article in Nautilus vividly illustrates, we’re also horrible at accepting randomness.

…when video games truly play by the rules, the player can feel cheated. Sid Meier, the designer of the computer game Civilization, in which players steer a nation through history, politics, and warfare, quickly learned to modify the game’s odds in order to redress this psychological wrinkle. Extensive play-testing revealed that a player who was told that he had a 33 percent chance of success in a battle but then failed to defeat his opponent three times in a row would become irate and incredulous

…So Meier altered the game to more closely match human cognitive biases; if your odds of winning a battle were 1 in 3, the game guaranteed that you’d win on the third attempt—a misrepresentation of true probability that nevertheless gave the illusion of fairness.

This notion of actual probability somehow being “unfair” is something to ponder. As is how the perversion of genuine probability can feel fair.

Are we more concerned with outcomes than opportunities?

It really is an excellent article and I recommend reading it all. It ranges from loaded dice found in ancient Egyptian tombs, to “pity timers” and faux improvement embedded in modern games, to gambling:

The results of any modern slot machine are based on arcane random-number generators in a computerized network, not on the fortunate conjunction of three wooden wheels. But losing to that sort of luck can be dispiriting. So gambling machines often employ the fiction of physical luck—by, say, making it look as if you just missed out on a king’s ransom as the final matching bar of gold or lemon reels to a stop just shy of a jackpot payout. This entices you to once more bet on odds that remain astronomical.

As always my emphasis.

Diversity compounds

…we find that black students randomly assigned to a black teacher in grades K-3 are 5 percentage points (7%) more likely to graduate from high school and 4 percentage points (13%) more likely to enroll in college than their peers in the same school who are not assigned a black teacher.

…We envision role model effects as information provision: black teachers provide a crucial signal that leads black students to update their beliefs about the returns to effort and what educational outcomes are possible.

This is from an intriguing working paper looking at natural experiments in Louisiana and North Carolina.

It fits with similar research into the importance and lasting effects of role models. They spur our dreams, change our attitudes, and shape our beliefs and expectations.

A superstar – especially one that is like you – can completely shatter self imposed limitations. Take this from a famous recent paper:

Girls are more likely to invent in a particular class if they grow up in an area with more women (but not men) who invent in that class… These findings suggest that there are many “lost Einsteins” – individuals who would have had highly impactful inventions had they been exposed to innovation in childhood – especially among women, minorities, and children from low-income families.

In the west we tend to focus on the individual. Often in zero-sum situations.

But when you add in the context of previous and future generations, we all have a stake in a visible diversity of success.

It will spur the next generation. It compounds.

Wouldn’t we all benefit from fewer “lost Einstein’s”?

As always my emphasis.

Private money for public goods

Almost every day for several years I would drag myself off a train at one of Sydney’s biggest train stations on the way to work.

If you’ve travelled, what’s striking about Australian train stations is how empty and tacked-on they feel. Apart from the odd shop and vending machine they are quite austere, functional and removed from the rest of life.

You don’t really get the supermarkets, vast food courts or even accomodation that are often found in Asia or Europe. Public transport hubs are places to pass through, not go to.

Australians are poorer for it.

After reading this paper about Tokyo railways I can’t help but think this is partly because Australian transport is mostly a government activity.

As government entities they are constrained in what they can do (lest they raise the ire of private competitors). This not only reduces their possible services and the resulting community and consumer surplus, it leaves them open to cuts and “efficiency dividends”.

The Tokyo railways on the other hand:

…remarkably for the twenty-first century, the private railways in Tokyo Metropolis operate at a profit…

…Government regulation of fares coupled with limited subsidies for railway operations pushed the private railways to innovate and diversify into a wide variety of related businesses, most notably real estate. Due to their long-term interest in the communities they built along their rail lines, the private railways provide valuable social benefits through public transportation while still pursuing profits.

Unlike Australian railways, the Tokyo railways are private and profit seeking. Coupled with heavy fare regulation, we have the happy accident of their expanding into other services and amenities as a way to maximise profit.

It is arguably one of the reasons for the building density around the railway lines in Tokyo. This increases walkability and helps integrate transport networks, making it far easier to get around without a car.

You now have a virtuous cycle even as the rest of us scramble to get people out of cars.

…private railways were able to survive and thrive by branching out into businesses closely connected with the railway industry, while private operators in Europe and North America slowly began to fail due to increased competition from the automobile…. On the Toyoko line, for example, Tokyu built high rise commercial centers at Shibuya station in Tokyo at one end and Sakuragicho station in Yokohama at the other, and opened its first department store near Shibuya in 1934. The corporation offered land at low rates to universities and schools in order for them to build campuses at intermediate stations along the way.

Tokyu is one of the most profitable railway operators in the country, with net profits of ¥58.72 billion (US$587 million, €441 million) on operating revenues of ¥263.7 billion (US$2.63 billion, €1.98 billion) in 2006 (Tokyu Corporation 2007). Real estate and transportation each bring in an equal share (33.5%) of net profit, with the rest coming from retail (20.2%) and other sources.

Of course, this could easily be an argument for complete government control. But I can’t imagine a purely government entity having the political might to get something like this done. Not without a whole lot else changing in society.

And anyway, without the ability to grow around the railway as Tokyo has done, you’re unlikely to replicate the effect. This isn’t a plan for established cities.

I guess this is more a call to be mindful of the extraordinary consumer benefit that can accrue if entrepreneurs are given the incentive to do so.

As always my emphasis.

Yet another benefit from educating girls

In many households, the oldest sister is one of the first family members to acquire any schooling, making her an important source of help with studies at home. Surveys show that only one out of five children who receive help with studies from a family member get it from a parent. When parents are not the ones helping with studies, the oldest sister fulfils that role 70% of the time. Given the low education of parents, and the oldest sister’s role as a childcare provider and tutor, one might expect oldest sister’s schooling to meaningfully impact younger sibling learning and development.

This is from research into the benefits of educated older sisters in Pakistan. Specifically in rural areas, where three quarters of mothers and two fifths of fathers were uneducated.

Javaeria Qureshi from the University of Illinois found that having an educated older sister increased a primary-aged brother’s years of schooling, ability to read and write, and add and count.

These oldest sisters’ schooling effects are the same order of magnitude as those for maternal schooling. Interestingly, increasing the oldest sister’s schooling has no impact on older brothers’ educational outcomes, indicating that the younger brothers are not benefitting merely from being around more educated family members or because their parents are increasing investments in all of their children’s education. Instead, the results suggest that younger brothers’ education improves because they have a more educated childcare provider, tutor and role model in their oldest sister. 

This is an important finding not only due to the number of children not being educated in Pakistan, but the gendered nature of education in many countries.

Women are often presumed to have a lower return on education due to a lesser propensity to work and support parents later etc. This can be especially important in resource constrained families even if there isn’t a prejudice against educating girls.

But as is so common, there’s stuff we haven’t or couldn’t measure, the world is complicated and factors interrelated. Beyond the individual benefit of educating a girl, it appears to be positive sum for the family.

As always my emphasis.