User:Niryhpr/Behind the QIDs

This entire report, abridged and summarized, fits onto a single page of an A4 poster.

Part 1 : The smaller the QID number, the more interesting

“ The "Q" in Wikidata Q identifier (Q43649390) was chosen by Wikidata (Q2013) founder Denny Vrandečić (Q18618629) as it is the first letter of his wife's name, Kamara Vrandečić (Q61768970). ” — Proc. ACM Hum.-Comput. Interact., Vol. 7, No. CSCW1, Article 84.

You see that string of numbers in every Wikidata item. But how do you interpret it? Do you ever look at the number behind the QID of a Wikidata item, pause, and wonder, “Why does this particular entity have this specific number?” Surprisingly, I think about that often.

I believe that the string of numbers behind every QID is not as "meaningless" as a UUID (universally unique identifier (Q195284)). Unlike a UUID, there is an inherent order behind QIDs, and from that order we can infer a lot of things. Especially about the collective activity of Wikidata users at a given time.

This is the reason why, whenever I craft my own SPARQL queries for a particular purpose, I really like to sort things based on their QIDs.

I made this query on December 11, 2025, at 22:32 UTC+7 (circa (Q5727902)). It is one of my favorite SPARQL (Q54871) queries to date. Before I explain what this query actually does, try to guess what kind of data it produces.

SELECT ?l ?lemma (xsd:integer(STRAFTER(STR(?l), "L")) AS ?numericId) WHERE {
  ?l a ontolex:LexicalEntry ;
     dct:language wd:Q13324 ;
     wikibase:lemma ?lemma .
}
ORDER BY ?numericId

“ The expression STR(?l) converts the lexeme entity into its full URI string, which looks like something such as http://www.wikidata.org/entity/L8348 . Then STRAFTER(..., "L") extracts the substring after the letter “L”, leaving just the numeric portion of the Lexeme ID. After that, xsd:integer(...) converts that string into an integer so it can be sorted numerically instead of lexicographically. Finally, ORDER BY ?numericId sorts all returned lexemes by that extracted numeric ID, effectively giving you lexemes in creation-order-like sequence based on their L-number. ”

In short, this query shows all lexemes in Minangkabau that are already stored in Wikidata Lexeme, sorted chronologically based on how early each lexeme was entered into the database.

Now, imagine this scenario: today is the grand opening of Wikidata Lexeme. The database is practically still empty, and you want to start entering lexeme data for your own language. From all the words you know in that language, which word would you add first, and why would you choose that particular word?

For me, these implications are interesting to ponder.

It turns out that the first Minangkabau lexeme added to Wikidata Lexeme is aia ("water").

Meanwhile, for English it is "first". For Indonesian, it is "satai/sate (L801)". For Betawi, it is enyak ("mom"). For Japanese, it is 子供 (kodomo, "children"). For Russian, it is the same as Minangkabau: вода ("water"). For Korean, it is 먹다 ("to eat"). For Arabic, it is حُلْو ("sweet").

My hypothesis is that when you are contributing to Wikidata, especially when creating a new item or lexeme, you are unconsciously “sorting” things in your mind. You tend to enter first the items that you care about deeply, or whatever happens to be present in your mind at that moment, or things that are simple yet very easy to recall.

In short, my hypothesis is that the smaller the QID number, the more interesting that item is relative to others.

Let’s analyze that "first lexeme ever" situation. Betawi and Japanese both suggest that human relationships are very important. Meanwhile, Indonesian, Russian, Korean, and Arabic all emphasize consumables. For Korean, the act of consuming itself is important. For Arabic, it is the sensation experienced during consumption that matters. Russian and Minangkabau align on water as the most important consumable, while Indonesian shows a surprisingly specific preference for a particular dish.

English, meanwhile, seems to transcend these categories, aiming for a more poetic framing when entering its first lexeme. Of course, the first lexeme has to be "first". Absolute cinema.

There are both "individual factors" and "communal factors" at play in this situation. Individually, people tend to prioritize adding items based on whatever they consider important. Meanwhile, collectively, Wikidata users are virtually "competing" with each other to add those items first.

These two unknown variables clash every day. The result is the QID numbering you see in Wikidata today.

Part 2 : The Big Bang of Wikidata

“ The Big Bang (Q323) is the scientific model describing the origin of the universe as an expansion from an extremely hot, dense initial state about 13.8 billion years ago. ”

Alright, so the main hypothesis is: “The smaller the QID number, the more interesting the item.” So let’s test it. What are Q1, Q2, Q3, and so on? After that, in what direction has Wikidata grown?

It all started on 30 October 2012, at 01:18 am UTC+7 precisely. A certain IP user “127.0.0.1” (wait, is that a localhost (Q153799)?!) entered several of the earliest Wikidata items to date.

Q1, universe (Q1). Q2, Earth (Q2). Q3, life (Q3). Q4, death (Q4). Q5, human (Q5).

Q6 to Q7 are skipped. They are already deleted from the database.

Q8 is happiness (Q8)..

Q9 to Q12 are skipped again. Q13 is triskaidekaphobia (Q13) (?!?!?)

“ “Triskaidekaphobia” means an irrational fear of the number 13. Etymologically, it comes from Ancient Greek. The first part, tris, means three, kai means and, and deka means “ten,” so triskaideka literally means thirteen. The second part, phobia, comes from phobos, meaning fear or panic. So the word is built quite directly as fear of thirteen. ”

Q13 is the "fear of thirteen"? Absolute cinema.

Q14 is skipped again. After that, user Denny entered the first continent ever in Wikidata: Africa (Q15). Specifically on 30 October 2012, 00:03 am UTC+7.

Wait, but Q1 is entered at 01:18? Why do items entered at 00:03 have a larger QID than items that were entered at 01:18!?

I guess the timestamping of the imported data works differently. Maybe it is timestamped only after all the data has been properly stored in the database. But I cannot really guess what is actually happening here. I need to dig into the source code to find out what is actually going on.

Anyway, just skip that mystery for now (I hope somebody figured it out later). The most important part is that this historic item creation defined the first "direction" in which Wikidata grows. What's next after "continent"? Well, of course: anything geographic. Nations, rivers, peaks, valleys, and so on.

Anyway, further analysis turns out to be quite tedious. I already prepared a half-baked draft over here and here.

Despite that incompleteness, I have already managed to uncover several interesting findings.

First, several of the earliest Wikidata users had already registered on the platform by October 27, 2012, before the first Wikidata item was completely imported via localhost on October 30, 2012.

Second, several of the earliest items created in Wikidata have no property statements at all. They consist of just an item label and, occasionally, a description.

Third, the entry for Indonesia was among the earliest items ever made. It was created on the exact same day as the localhost import (October 30, 2012). User PiRSquared17 created the item at 8:52 AM GMT+7, and it received the identifier Q252. Meanwhile, the first country on Wikidata was Q16 (Canada), closely followed by Q17 (Japan), both of which were created by user Aude.

From this starting point, we can classify the country items into two groups: those created before Indonesia and those created after Indonesia.

Here is a sample of the "Before Indonesia" group: Q17 Japan, Q28 Hungary, Q29 Spain, Q30 US, Q31 Belgium, Q34 Sweden, Q35 Denmark, Q36 Poland, Q38 Italy, Q39 Switzerland, Q40 Austria, Q41 Greece, Q43 Turkey, Q45 Portugal, Q55 Netherlands, Q96 Mexico, Q114 Kenya, Q142 France, Q145 UK, Q148 China, Q155 Brazil, Q159 Russia, Q183 Germany, Q212 Ukraine, Q219 Bulgaria, Q225 Bosnia and Herzegovina, Q230 Georgia, Q241 Cuba.

Here is a sample of the "After Indonesia" group: Q258 South Africa, Q262 Algeria, Q298 Chile, Q334 Singapore, Q403 Serbia, Q408 Australia, Q414 Argentina, Q419 Peru, Q423 North Korea, Q424 Cambodia, Q458 European Union, Q574 Timor-Leste, Q664 New Zealand, Q668 India, Q691 Papua New Guinea, Q695 Palau, Q712 Fiji, Q717 Venezuela, Q730 Suriname, Q739 Colombia, Q794 Iran, Q796 Iraq, Q805 Yemen, Q819 Laos, Q833 Malaysia, Q838 Myanmar, Q837 Nepal, Q843 Pakistan, Q851 Saudi Arabia, Q854 Sri Lanka, Q865 Taiwan, Q869 Thailand, Q869 UEA, Q881 Vietnam, Q884 South Korea, Q902 Bangladesh, Q921 Brunei, Q924 Tanzania, Q928 Philippines, Q1019 Madagascar, Q1028 Morocco, Q1033 Nigeria.

As for "what is the specific pattern within this classification?", we leave that as an exercise for the reader.

Fourth, one of the earliest Wikidata users from Indonesia is user Wagino 20100516.

The very next day after the localhost import—on October 31, 2012, at 7:52 PM UTC+7, he added item Q1795 "Asahan" with the description "Asahan Regency is a regency in North Sumatra".

Right after this began one of the earliest projects in the history of Wikidata Indonesia: importing all the geographical names of Indonesia into the Wikidata database, including cities, villages, districts, subdistricts, provinces, islands, mountains, rivers, and so on.

In contrast with PiRSquared17's earlier edits, which simply added a list of country data to Wikidata, this entry marks the first recorded Wikidata item creation by an Indonesian citizen. Therefore, if Wikidata Indonesia ever wants to pick an alternative date to celebrate their birthday, October 31, 2012, at 7:52 PM WIB would be an excellent choice.

Then, the following night (relative to Jakarta local time, UTC+7) on November 1, 2012, at 2:55 AM, one of the earliest Wikidata items for an Indonesian individual was entered.

It was created not by an Indonesian user, but by a French-speaking Wikidata user. In the description field, that user wrote: "Homme politique indonésien (1966-".

Guess who?

Q3745 Ahmad Heryawan.

In 2012, he was serving as the Governor of West Java. He had been in that position since 2008, and 2012 was still within his first term (2008–2013). At that time, he was already preparing for re-election in the 2013 West Java gubernatorial race, meaning he was acting as an incumbent candidate while still holding the governor's office.

Why a French-speaking user was interested in the Governor of West Java remains a mystery to me.

Fifth, on that same day, I found one of the earliest traces of bot editing.

On November 1, 2012, at 5:09 AM early morning Jakarta time, MerlIwBot created Q4315 "Subak" with the comment: "Robot: Adding id,de,ja,fr,ru,he,sh,ka,w,hr". The bot was probably using the titles of existing Wikipedia articles, which were readily accessible via bulk database queries, and mass-importing them directly into Wikidata.

Sixth, the next day on November 2, 2012, we see a return to organic, manual Wikidata item creation by various human users across a diverse range of topics. At 2:25 AM, user Reedy added Q6498 "CASA/IPTN CN-235"—one of the most famous "Made in Indonesia" aircraft. At 4:23 AM, user Sannita added Q6940 "UTC+07:00" with the description "Identifier for a time offset from UTC of +7".

The next day, on November 3, 2012, at 3:51 AM, user Sumone10154 added Q9240 "Indonesian" (language).

Seventh, at this point, Wikidata is growing so quickly in both depth and breadth. So I need a better alternative method of analysis to continue this investigation.

Part 3 : MyWikidataList

Not all Wikidata items are created equal, at least for me, personally. To me, there are some Wikidata items that I love more than all the rest.

Starting from the hypothesis that "the smaller the QID, the more interesting the Wikidata item is, because they probably prioritized things they believed were important when first entering them into the database," let's expand this idea further.

What is your favorite Wikidata item, and why?

I decided to run this experiment on myself to find out what my favorite Wikidata item is and why. That is exactly why I created this subpage.

But how do I choose which Wikidata item is my favorite?

Can we produce a list of all Wikidata items so I can inspect them one by one and handpick a favorite?

To do this, I modified the query "all Minangkabau lexemes, sorted by their QIDs" by changing the "Minangkabau lexemes" part to whatever category I wanted to explore.

After running those queries in the Wikidata Query Service (WDQS), I took notes on how many items were returned for each specific category. Then, I manually sorted the categories based on their item counts.

For Indonesia-related Wikidata items, I discovered that "human" is one of the largest categories. A total of 107,228 Indonesian humans were successfully extracted from WDQS at the time of the query, and that number is likely increasing as we speak.

Are all of those humans notable enough to be included in Wikidata? Let's save that question for another day.

The next largest category is "desa" (literally meaning "village"). In the Indonesian language, the term "desa" can refer to two different concepts. First, the exact same sense as the English word "village," and second, a rigidly defined administrative territory recognized by the government.

I assume this massive count is simply a result of one of the earliest Wikidata Indonesia projects to import all geographical data. It was likely mass-imported from official government sources.

The next largest category is "politician," which is a subset of the first "human" category. It totals 52,194 items.

Wow. Almost 50% of the humans in Indonesian Wikidata are politicians.

My guess is that this data was also imported from an external database. Likely legislative election results. That would provide plenty of data for a mass import.

Are all of those politicians notable enough to be included in Wikidata? Let's save that question for the future as well.

The next groups are other subsets of either the human or geographical categories: rivers (42,116), "kelurahan" subdistricts (8,881), and actors (3,121).

Finally, past those, we see some really fresh, diverse topics : films (2,381), badminton events (1,348), languages (1,329), railway stations (1,084), association football clubs (884), businesses (588), and musical groups (451). I really love this section because each category utilizes a unique set of domain-specific Wikidata properties, which is excellent for expanding the breadth of Wikidata's coverage.

The remaining categories are too small to be worth reporting.

Keep in mind that this entire list is incomplete. I am just randomly brute-forcing my way through guessing these categories rather than systematically examining the entire Wikidata database.

Because of this, there might be even larger undiscovered categories out there. This list serves merely as a sample. A preliminary analysis to get a rough grasp of the overall situation.