Minh’s Notes

A well-read op-ed in this week’s Wikipedia Signpost lists off a number of well-known Wikipedia hoaxes over the years. One of these, about a supposed mayor of China, persisted on the site for years and became fodder for a Harvard admonition against using Wikipedia in academic research.

In 2008, during my internship at Microsoft, I heard through the grapevine about someone who wrote a Wikipedia article on himself, falsely claiming to be mayor of a city in China. Despite not knowing this hoaxer’s identity, it wasn’t difficult to come across the hoax. Few Chinese mayors had their own articles at the time, and this one didn’t read like a typical stub article. It was filled with personal details, non-résumé details, the kind of stuff editors grudgingly add after running out of ideas, just to round out an otherwise complete article. A quick Google search showed that there was a mayor, and it wasn’t the subject of the article.

So in the midst of an already productive night of editing, I flagged the article for deletion as a hoax and noted my concerns and evidence on the talk page.

The next morning, a total stranger stormed into my office, demanding that I take back my comments and publicly vouch for the article’s authenticity. He came prepared, pointing to a printout of my Wikipedia user page, then to a link to this blog, and finally to a comment my friend had recently left on this blog that hinted at my connection to Microsoft. He chided me for not properly maintaining my anonymity like he had.

Yeah, whatever. It’s not like I ever tried to hide my identity on Wikipedia. But who was this guy? For all I knew, someone interning in HR? I had never been challenged in real life about an action I took on Wikipedia. So I played along, feigning shock, and hammered out an agreement that he would take down the deletion notice himself and replace it with a source. My plan would be familiar with anyone fighting vandals or trolls: wait until they lose interest, then quietly wipe away every trace of their work.

In accordance with the agreement, an anonymous user soon removed the deletion notice and provided a link to a document that supposedly backed up the entire article. The page sat behind a paywall on a site like Newsbank or ProQuest, so an ordinary user wouldn’t be able to verify the source. I had a subscription through Stanford, but I’d have to wait until I returned to campus that fall.

That page has remained in my overflowing “to do” bookmarks folder ever since, a victim of my world famous procrastination. Meanwhile, the hoaxer apparently tipped off his alma mater’s writing center to the hoax and probably received fine accolades from the school library for his work. The article was finally deleted this past November, in response to Harvard’s guideline.

So no, you can’t trust Wikipedia for your academic research, because your classmates are perpetrating lies there, the bystanders are too lazy to act, and your school is chuckling hysterically.

Periodically, I still receive inquiries about the Atmosphere Community Server, which powered real-time, multi-user chat inside the 3D “worlds” of Adobe Atmosphere. Long before they pulled the plug on their copy of the server and discontinued Atmosphere development, Adobe released the server software under a very permissive open source license. However, obtaining the software has always been a challenge: Adobe no longer accepts signups to host the server, and it no longer seemed to be available for download anywhere online.

A few years ago, after searching high and low for a chat server to hook my world up to, I finally got my hands on the server source code. Joe De Costa was one of the few enterprising hobbyists who ran their own chat server. He meticulously documented the steps to setting it up on his Linux computer, but reading his tutorial easily dissuaded me from setting up my own copy.

The source code as it exists today is something of a game of telephone: as successive hobbyists have handed it down over the years, they’ve each jury-rigged the software to run on their Windows and Linux computers, not necessarily keeping all the features intact. The changes have all the hallmarks of guess-and-check, which is understandable given the disconnect between the server’s systems-style, multithreaded C and the DHTML-era JavaScript that most world builders were familiar with.

Despite my initial apprehension, curiosity got the better of me, and I sat down one weekend to get the thing up and running on my MacBook Pro. The result is an Xcode project that builds and runs in a 32-bit Mac OS X environment, with far fewer caveats than the package I received. You can download either package here. Of course, please respect Adobe’s intellectual property, which should be easy given their downright liberal licensing terms.

I’m under no illusion that Atmosphere will one day return to its former prominence. After all, the industry has moved on and there’s only so much a ragtag band of hobbyists can do to revive a dead, closed-source platform. But Atmosphere has a allure for me, as I suspect it does for the people who occasionally contact me about it. It was the one 3D platform that was never exactly a game, or a shopping experience, or a chat room, or an HCI research project. It was open-ended and decentralized, like the Web. Maybe that’s why it flopped. Then again, maybe another, more successful Atmosphere will come our way in the future. We’ll be ready to pick up where we left off in 2004.

I have a startling admission to make: I never really used the Encyclopædia Britannica. My classmates and I instead feasted on the more accessible World Book, in all its photo-bound glory. At home, the only encyclopedias my family could budget for were the first two volumes of Grolier’s Encyclopedia of Knowledge, which randomly went on sale at Kroger one weekend: A–Ano and Ano–Bas. They were beautifully illustrated and typeset, but as you’d imagine, my school report on Alabama went on a bit longer than the one on Virginia. Later on, Encarta filled in the rest of the alphabet. The lack of a particular upscale brand of encyclopedia at a decidedly non-elementary reading level never registered as a problem.

So with Britannica’s print demise this week, I have a hard time understanding how an empty Britannica shelf would be the greatest of our concerns if the national power grid went kaput. It isn’t as if encyclopedias – even print encyclopedias – are done for. The ones that found their niche in the education market are doing fine. I also don’t buy the argument that our society’s reliance on Internet sources will deprive rural public libraries in the U.S. After all, a computer is half the price of a full Britannica set, yet the former’s value is certainly greater. Can Britannica walk you through algebra problems the way Wolfram Alpha does? Would you type on it?

The majority of nostalgic commentators seem to pin the blame for Britannica’s exit on Wikipedia and find fault with its directness. They remember sticking their fingers inside a huge book to keep their place across cross-references, and fascinating entries that happen to lie on the same page stealing their attention. And they lament that a search box can’t offer any of that.

To solve that problem, I’ve written Serendipity, a script that allows registered users of Wikipedia to bring the true encyclopedia experience to the site. Install it, and you’ll quickly notice that whatever you look up is preceded and followed by entries that just happen to fall next to it in alphabetical order. Because this model is so essential to cognitive development and intellectual curiosity, the script also disables hyperlinks and other aids to non-linear navigation.

Of course, the script is totally tongue-in-cheek. But I hope you’ll like it – for all of three minutes. In a few months’ time, we’ll mostly forget about the print Britannica just as we’ve mostly forgotten about card catalogs. Indeed, anyone younger than me won’t even know what they are. And we as a society will be alright. Digression is important for learning, but it doesn’t entail alphabetical order.

Well into my freshman year of high school, extreme boredom led me to give in and start a “blog-in-kind”. Ten years later, here we are.

(Oh, typographers and their silly terminology.)

One of the crazier projects I’ve begun recently has been to transcribe a 500-page trilingual dictionary from the 17^th century for Wikisource, the wiki document archive affiliated with Wikipedia. The dictionary in question, Alexandre de Rhodes’s Dictionarium Annamiticum Lusitanum et Latinum, defined what would come to be known as the Vietnamese alphabet.

In one of the first attempts to apply a European alphabet to an Asian, tonal language, de Rhodes had to innovate somewhat. For the Annamese-Portuguese-Latin dictionary, he derived two additional letters and invented three completely original diacritical marks. He also mixed and matched from at least a few European languages in what his contemporaries must have considered cutting-edge phonetics. Yet, in the midst of the Renaissance, typographical flourishes – and typos – also abound in de Rhodes’ works, which were published by the Propaganda Fide, the missionary organ of the Catholic Church in Rome.

The result is a mix of traditional details and prescient quirks. De Rhodes’ novel, 17^th-century alphabet continues to the present day largely intact. But some of the letters and diacritics he invented have gone extinct or now represent different sounds. The rich display of ligatures and swashes, then common in printed text, appear only tongue-in-cheek or erroneously today. In particular, de Rhodes’ type designer sometimes kept the tittle on his accented “i”, depending on the diacritical mark.

As in most historical literature, these typographical details are easily lost when the text gets digitized. So to help preserve these anomalies, I adapted an existing, well-regarded medievalist typeface, Peter S. Baker’s Junicode, for archaic Vietnamese texts like de Rhodes’. I beefed up support for modern Vietnamese; trimmed the font down to just the characters in Vietnamese, Portuguese, and Latin; and added de Rhodes’ novel characters. The result is probably the first font to support the letter being proposed for inclusion in the Unicode standard as “B with flourish”. It looks as though someone meant to write a “b” but midway through started drawing an “@” instead.

The font is named Đắc Lộ, the traditional rendering of de Rhodes’ name in Vietnamese. Even if you can’t read Vietnamese, the font’s download page is worth a look. Intent on replicating the old-world book feel, I stuck to typography, rather than imagery, to make the case for a rich but sturdy typeface. There are no images. Đắc Lộ is embedded as a Web font, and (in Firefox and Camino) you can barely discern a completely textual table of glyphs on the back side of the page, thanks to some shameless CSS trickery.

With typography out of the way, it’s time to continue transcribing the dictionary. I’ve gotten to the Bs so far.

Thanks to Peter S. Baker for giving his font a generous license. My efforts in this space are trivial compared to his.