User:Leaderboard/WMF Quiz 2024/Solutions

Spoiler: contains solutions

Overall notes

There were a couple of issues that did affect this quiz:

Mobile users seemed to be unable to see questions with a comprehension text type. This was partially mitigated by telling the data before the timer started, but two questions had to be skipped.
For some reason, the CEE group ended up putting a quiz at the same time as mine (i.e, unannounced collision), which appeared to negatively impact the participation of this quiz and was very frustrating to see. This was unfortunately a negative aspect, as this is one event that cannot be made up after the quiz has concluded.

There were also some complaints from participants that appeared to struggle with the more demanding questions in this quiz - it appears that this was not made clear enough as very few appeared to read the specification document before the quiz. In a future iteration (assuming that it even gets accepted), this will be made clearer. To be clear, this is expected - I did not expect anyone to get all the questions correct and for this iteration, getting more than 80% the points (with the two skipped questions) would be evident of an excellent participant (and ~62% of a very good candidate). Similarly, a few complained of some of the questions testing technical skills - this was again mentioned in the specification. The whole goal of this quiz was to allow weaker candidates to get some of the points, while stretching the more able ones - this will be emphasised further in a future iteration.

Question 1

The answer, obviously, is Wikispecies. This was a friendly starter to the quiz, with the vast majority of this getting it right.

Question 2

The answers are yellow and sunflower, respectively (see right). This one was done OK, but cracks started to emerge amongst some of the weaker candidates, who got it completely wrong or left it blank. Partially correct answers to this were very rare.

Question 3

The correct answer is B: The code runs and returns a TLE (time limited exceeded) message. In the past, the answer was actually A (i.e, it would run and return nothing), but changes were made recently. Option C (which was quite popular) is nonsensical - the system does not have "detection" capabilities; option D would be a waste of resources, and option E is not normal practice.

This one was problematic because those on mobile could not apparently see the code. Despite this being explained to participants (i.e, we're dealing with an infinite loop), this was one that started to challenge candidates as a whole - all the options were chosen by a significant proportion of candidates.

Question 4

The correct answer is E - this was one case where the developer explicitly licensed the image to be free-use and a VRT ticket relating to this can be seen on the description.

This one was harder than anticipated because of the trap option D ("It is free-use because the game itself has been released under CC BY-SA 4.0, and hence a screenshot from it must also be free-use"). The problem is that this is correct if we know that the game is open source, but that isn't actually the case. This is a rare example where the developer has agreed to licence an image as free-use, and this is something more games should do, in my view.

Question 5

This was reasonably well answered (amongst those who did attempt it) - the correct answer is option A ("Page A goes to page B, and stops there."). This is because on page B, the redirect flag (which you can see in the link) would be set to "no", preventing the page from going back to A.

Question 6

The correct option is E ("Internet Explorer complains that the page cannot be displayed"). The reason is actually due to protocol errors - Wikipedia requires 256-bit encryption which IE5 cannot (in fact, no version of IE would work on Windows XP because for some reason, it's limited to 128-bit - this includes IE8 when the same configuration would work on Windows Vista/7).

This was the first question I intended to be challenging, and it did end up being the case - this was discriminating and very few managed to get this right. Option D ("The main page appears, but with formatting errors") was very common (and admittedly a sensible one, because that's what would normally happen if the site was capable of being served over HTTP or relatively less-secure HTTPS, like with Google).

Question 7

The "true" options are "It has 16 members in the committee when fully seated" and "It can order the closure of wikis if necessary", with the others being false.

It was surprising that no one managed to get all correct, though partial marking turned out to be very helpful as many managed to get some of the points. Clearly not many were familiar with the U4C charter, which was what this recall question was testing.

Question 8

This was another question that caused a lot of problems, with many struggling to understand what was required for them. Many seemed blindsided with the ramping up of difficulty as the quiz progressed and were not expecting a question testing (simple) regex.

The correct options are B (".S.") and D ("(S)"). The key idea is that both of these options showcase the key issue - that it captures the bad word S as part of a larger word (so if S = xxx, the string yxxxy would be caught). Option A (which is just S) requires the exact word, C requires the exact word fSf, and option E requires three copies of S in the word.

Update: this solution is flawed. A couple of users correctly pointed out (in the words of one of them) that "putting it in a regex group doesn't make it stop matching the full string either", and I made a regex error as a result. This means that option A would work (along with B and D) for instance, as the capturing still occurs, and a different regex will need to be used instead if this capturing shouldn't occur. Most users did not get anywhere close to this question so the impact would be limited, but it would have affected the more able participants. Unfortunately, scoring cannot be changed as I was notified of the error more than a month after the quiz took place.

Question 9

The correct answers are Europe -> Asia -> Africa -> South America -> Oceania [6 -> 4 -> 2 -> 1 -> 0].

This question had a variety of scores, with some getting all correct, many getting some right, and a few answers being completely wrong. Again, partial marking helped a lot of participants here.

Question 10

This was skipped

The expected way to do this one is to look carefully at the code, which says that a "newbie" (i.e, non-autoconfirmed user) can move two pages in two minutes, or one page per minute or $60\times 24$ pages per day (which is 1440). After they are autoconfirmed, this rises to 8 pages per minute, or $8\times 60\times 24$ pages per day.

Now, it would take two more days for the user to become autoconfirmed, so the user can move only $2\times 1440$ pages in two days. The user can move $5\times 8\times 1440=40\times 1440$ pages in the remaining five days, which means that in total the user can move $42\times 1440$ pages in total. Hence p = 42.

Question 11

The correct option is (obviously) C, administrator. This was a relief for many participants for obvious reasons.

Question 12

The extensions that can be uploaded are GIF, PNG, MP3, MPEG and PDF.

This question was designed in a way that allowed weaker candidates to get the "obvious" options, while reserving full points for the strongest candidates (and one managed to get all the way to the end). The tricky parts tended to be relating to MP3 (which you can upload, as the question explicitly says that there are no user restrictions), MP4 (which you cannot upload as it isn't free) and EXE/DOCX/others.

Question 13

This was skipped

The expected way in this case is to visualise the graph - there's one on Wikidata itself somewhere.

In the pre-Wikidata implementation, each of the k languages would have k - 1 links to each of the other language pages. Since the links are bidirectional, we divide this by 2, resulting in the answer ${\frac {k(k-1)}{2}}$ .
In the Wikidata implementation, the central Wikibase server would have one link to each page, or k links in total. That makes the correct answer k in this case.

Question 14

The correct options are Poland, Armenia and Frankfurt.

This question was testing recall of something that was covered in the opening ceremony of Wikimania 2024 (the first blank), with the other two requiring outside knowledge (or at least a curiosity to read it after the ceremony!). Invariably, the first option was correct (amongst those that did not get a 0), with the other two being a toss-up. No one got all three correct.

Question 15

The correct option is B: "Only users that are “autoconfirmed” can create any article on mainspace" - about 2/3 of attempts on this was correct. Incorrect answers were all over the place in terms of options.

Question 16

The correct options are Wikitech (SUL migration will soon be in progress but isn't the case as of writing) and VoteWiki (which has always required a separate account).

This was surprisingly poorly done, with less than 10% of attempts being correct (and was not of a partial-marking type unfortunately), Very few seemed to have a clue about VoteWiki, and many chose Foundation Wiki (which used to require a separate account but was migrated to SUL some time back).