Sweden just claimed the world’s second-largest Wikipedia | Hacker News Reader

Sweden just claimed the world’s second-largest Wikipedia

washingtonpost.com

68 pointslelf44 comments

I'm not sure I'd say it's the 2nd-largest. Second-most articles, yes, which is a different metric. Total size of an encyclopedia has to include not only the number of headings, but how much text is under them! For traditional paper encyclopedias, something like word-count is typically used, though that is tricky to use across languages, since languages have different notions of what constitutes a word (and different information rates, if you want to get into that). Something like "compressed size of the database" might be an approximation for encyclopedia size that normalizes for different languages' use of words & UTF-8 bytes.

I agree, I'm a swede myself but almost always use the English version because the Swedish articles usually contains very little information.

It's exactly the same thing for me in Dutch. There are a lot of articles, sure, but the content is woeful. And there aren't even bots that artificially inflate the numbers there.

I agree for general articles, but for an article that has regional specificity, I'll look up the article in the local language and use an online translator.

Yeah, I do that with French and German especially. For very well-known people and places, the English article is usually good, but if you want something only moderately known, like a French writer who is historically important, but not Voltaire-level famous, there's a good chance the other (in this case French) article will be more complete. Also true for Japanese Wikipedia articles on Japanese people and places. But Google-translated Japanese->English is hard to read, so I unfortunately rarely consult those.

On the other hand, some of the smaller-language Wikipedias can be quite one-sided and nationalist when it comes to local topics, especially disputed history.

I do the same, but for most subjects the english article is enough. I rarely need more detailed information than that, but if I do, and I need it, I will use a translator too.

Good point.

From an analysis of parallel UN documents, it seems that character counts are about 10% higher in Spanish/French/Russian than English. The same documents are translated to Chinese with about 1/4 the total characters.

My instinct is that character count is a pretty good approximation of size for European languages, but a big adjustment would need to be made for some Asian languages where a character is more like a syllable in european languages.

http://www.uncorpora.org/Rafalovitch_Dale_MT_Summit_2009.pdf

The number of articles is a terrible metric of the quantity of content of a Wikipedia. A far better way to measure this is to look at the size of the data dumps of http://download.wikimedia.org (compressed so as to normalize differences in information density between languages).

(Shameless plug: I tried to do that three years ago http://a3nm.net/blog/wikimedia_projects_by_size.html -- it could probably be done again today, though the code would probably need to be refreshed a bit.)

Before Wikipedia there was a Swedish collaborative encyclopedia called susning.nu [1]. It eventually shut down due to problems with vandalism and in the meantime sv.wikipedia had grown larger. I suspect a lot of articles were migrated to wiki from susning, or at the very least the community did. Interestingly, susning roughly translates to grasp, as in a short format understanding.

"Because of this, Susning grew and became Sweden's biggest and the world's next biggest wiki." [1]

[1] http://en.wikipedia.org/wiki/Susning.nu

I remember we relied a lot on it in the early 2000's, as Wikipedia at that time contained more general information, and susning.nu was useful for more local (to Sweden at least) information.

Actually, not that much from Susning could be migrated due to license, rights issues.

The creator of Susning, Lars Aronsson is very active on Swedish Wikipedia. As far as I understood him, the content he knew the rights to (his own) could be migrated. But since the wiki is a collaborated effort and Susning had such licensing terms most stuff could not be attributed and is stuck in Susning.

> And it certainly doesn’t know that there are only 9 million people around to read its output, versus the 75 million who speak French, or the 78 million who speak German.

Minor nitpick: https://en.wikipedia.org/wiki/List_of_languages_by_number_of...

FR: 74m

DE: 89m

That's assuming Wiki P is correct-ish and that page is not some Teuto[+]-bot generated nonsense

[+] https://en.wiktionary.org/wiki/Appendix:English_nationality_...

There are over 80m people living in Germany which is already more than 78m, and this is totally ignoring German speakers in Austria, Switzerland, Belgium, Denmark, Italy and Poland. I don't know how they came to those numbers.

You forgot France, Lichtenstein and Namibia.

Why actually spend all this effort on localized Wikipedia articles? Like someone commented, many people read the English Wikipedia because it contains more info. At least I do, I know most peers do, and a Swedish commenter (fiskpinne) just mentioned it. Sure, local articles are good to have for many basic subjects, but when you really want to get into something you probably know English.

People shouldn't want to read articles like the one about cryptography (just picking something technical at random) in a non-English language beyond what the word means and a general overview. There are so many English resources and so many people able to read English if you'd contribute resources, learning it should be encouraged rather than translating everything just because we can.

>Why actually spend all this effort on localized Wikipedia articles?

Because most people can't read English.

If the world's best encyclopaedia was only available in Swahili, would you still be asking "Why actually spend all this effort on localised articles?", assuming of course that you're not a Swahili speaker.

If almost everyone in the developed world spoke Swahili, it seems pretty reasonable to say that.

What? Wikipedia is for everyone. Most of the worlds population does not speak english.

As a swede I use both swedish and english Wikipedia. It depends on what I'm looking at. Swedish Wikipedia contains more in depth content on a lot of topics.

> Because most people can't read English.

That's a good argument for some languages, but for Swedish in particular it's less strong, because the number of people who can read Swedish but not read English is fairly small, and rapidly declining. That doesn't mean there's no point to a Swedish Wikipedia, but whatever point there is can't really be because Swedes are unable to read English.

That's also true of Danish, which is one reason I believe the Danish Wikipedia has been languishing. There are many Danish people who are Wikipedians, but many of them do more English editing than Danish editing. The Swedish Wikipedia seems to be more active in comparison. It would be interesting to understand why; is it just the population difference (9.5m vs. 5.5m)?

Why is this attributed to "Sweden" rather than to "Swedes" or "Swedish speakers"?

* Swedish is the first language of about 5% of the population of Finland

* Swedish is supposedly mutually comprehensible with Norwegian

* Many minority languages are governmentally recognized and spoken by significant minorities in Sweden, including Saami and Finnish

There should be name for this fallacy of equating language and nationality.

(edit: newlines)

This article poses a real question whether computer generated articles are suitable for encyclopedia consumed by humans.

Speaking of this, I would love to see Wikipedia in machine-readable language. Something like CYC project.

Edit: I know of WikiData, but there should be an unstructured analog to that.

As long as the content of those articles is relevant and accurate, I don't see how that would impact on the suitability of the articles in question.

Does it really matter if a bot or a human has created an article as long as it contains facts and not thorough analysis?

Disclaimer: I haven't read any of the bot generated articles (at least, I don't think so).

I agree with your point. It's generally better to have some article that no article. But Wikipedia is also supposed to be readable by humans, to provide insight and understanding. I am not sure that mere formatted collection of data, as opposed to a real article, does that. And I believe there are many Wikipedia editors who feel the same.

Though, wouldn't it be more efficient to put the data into some machine readable form, if they were already processed by the machines anyway? This would then allow anybody to create algorithm to extract information (meaning) from the data.

So I think there are also good reasons to keep the human and machine created articles separate.

It's not so much a formatted collection of data as it is like an excerpt from an encyclopedia. It's sparse and the articles are pretty similar to each other but it has proper sources, and someone might actually find the information useful. And it could serve as a starting point to improve by some enterprising Swedish biologist.

If you want to read any of the bot generated articles, just try http://sv.wikipedia.org/wiki/Special:Slumpsida

I tried it 10 times and got 6 bot articles.

You're better off reading the English version rather than the poorly translated Swedish one imo.

If I understand the articles correctly, these are not Wikipedia entries that were translated from English into Swedish. These are entries created for the Swedish Wikipedia from original source material written in Swedish. The article mentions plans for creating articles for authors that do not have entries in the English Wikipedia. This means that the Swedish Wikipedia has a large number of entries on topics that do not have entries in the English Wikipedia. And will have even more in the future.

It's something like this: https://sv.wikipedia.org/wiki/Lepechiniella_persica. Basically, if you click "Random article" (https://sv.wikipedia.org/wiki/Special:Slumpsida) you have a about a 50/50 chance of hitting an article by the bot.

Some of them even have pictures but most are pretty sparse. In my opinion, it's not that bad to have these articles but they should be improved.

Of course it is, because it is full of GARBAGE made by this guy

http://online.wsj.com/articles/for-this-author-10-000-wikipe...

His stupid bot pollutes wiki with useless crap like this:

https://sv.wikipedia.org/wiki/Lepechiniella_persica

Hehe, that guy was my physics professor. He is a genious. Had a majestic beard as well. His stories and teaching methods reminded me of Carl Sagans style.

But his bot is quite annoying yes.

The first my reaction is to say it's pretty cool or something, but actually it isn't if I look at it more closely. You see, I don't really care about lots of information gathered on the website with one domain. In fact, ideally I would wish for exactly the opposite: if some information can be easily classified, then it should. What I mean is that adding every new article to wikipedia has its own drawbacks as well as value. Recently I was thinking about how cool it would be to have wikipedia accessible offline on some portable device — and it's nothing new, I remember edited versions of wikipedia for mobile devices before smartphones become popular, but this sort of things isn't very useful in many ways. On the other hand, realistically I would assume that all content I will need to look up in my portable wiki easily fits on modern storage cards even with some media-data. Even more than that: many articles become useless to me if images are removed. The only real problem is that I don't know what I will look up and it's very hard to define what's useful.

Anyway, perfect layout for animal articles is significantly different from perfect layout for math articles, so instead of populating wikipedia with rather obscure non-verified by humans information the really useful work would be making easy-to-use thematic resources as the existing ones made by academic people usually are totally unusable. That's where structured information should go in the first place.

However bots on wikipedia often do some useful work, what really makes wikipedia useful is mindful human work. Media-wiki engine isn't perfect for structuring information and is useful specifically for free-form, unstructured data populated by humans. I don't really see what is the use of populating wikipedia with thousands of auto-generated texts.

What a dreadful chart

When a chart is harder to interpret than a simple table with the same information, you know something went wrong. Or at least you should.

So they're fine with a bot filling in obscure information that no person decided was worthwhile to add manually, but if you add something manually it can easily be deemed too obscure to be useful? In a strange way, this policy is a logical extension of the actual (not the claimed) nature of deletionism.

Likely exclusively due to this guy:

https://news.ycombinator.com/item?id=8033600

Who runs a bot that makes millions of shitty stub articles.

Which is most of what this article is about...

I thought it was fitting to link to the HN discussion, but actually linked it before reading TFA. Shame on me.

Isn't it about quality, not quantity?)

I wonder if there is an alternative to Wikipedia that doesn't suffer from the same insular type of "Self-Appointed Guardians of the Galaxy"?

I mean one with a more sophisticated system for decision making rather than just bureaucracy, fiefdom, politics, and bribery?

If you are looking for a crowdsourced news archive and biography information (which accounts for around half of Wikipedia's pages) you can check out my site, Newslines [1], which fixes many of Wikipedia's software, data presentation, and policy issues. We do this by 1) having a simplified data structure based on news events 2) having a non-combative editorial approval system and 3) paying our writers to post.

[1] http://newslines.org/

If they want to get past the English side they can get a good start by translating all the anime and music related pages, what good are three million or more articles in english when a good number are about individual songs or anime characters?

“Wikipedia is written by affluent white nerds, in languages only affluent white nerds know"

so his solution to this problem was to expand the volume of content for a language spoken almost exclusively by white nerdy people

And the main page of Wikipedia is not updated, but: https://meta.wikimedia.org/wiki/List_of_Wikipedias

It is now, that's what makes it so awesome!

Reply on news.ycombinator.com