This year Microsoft banned April Fools, which deprived the internet of their unique sense of humor as illustrated by Clippy and Windows 9. Stack Overflow and Stack Exchange users were not so lucky. Inspired by Captain Marvel, Space Jam and our own misspent youth, we inflicted a 90s theme on the network.
Reception
Judging solely by messages from former colleagues, family and acquaintances of employees, the theme was a hit. The cursor trail was especially popular. Twitter was generally into it, though a significant percentage of responses were annoyed with the distraction and a few noted they had to leave the page. Reddit had roughly the same reaction and Hacker News was a big old "meh".
Responses to my MSE announcement (+347/-48) and my MSO announcement (+863/-22) were generally pleasant to read. Several people suggested features that we did in fact include such as an old-school view counter (it was just our existing question view count restyled) and "Best viewed in Netscape" (way down at the bottom of the page). There were suggestions about how to make the page more authentic (more tables and frames) and the inevitable digging for Easter eggs (such as the ASCII art in the source). Some people thought the site was hacked and others seemed genuinely unsure if it was intended as a joke. Due to timing more than a few people wondered if the idea was from our next CEO.
Some people seemed quite angry with the page as they didn't want the distractions while trying to get their work done. In fairness, we launched with a bug that required users to turn off the feature on each page. While we did test the feature, it probably makes sense to reuse the toggle from previous years when possible to avoid this sort of error. (It should also be a network-account-wide setting.) I also suspect the sheer quantity of distractions on the page made finding the toggles close to impossible for some. If you are in the middle of solving some other problem and you have to try to solve our toggle puzzle, I can see being pretty frustrated with the whole thing.
This isn't our first rodeo. Some people will hate these events (either in principle or in the particular way we handle them). We still think April Fools is a worthwhile tradition. There's no real way to measure the unhappiness caused compared to the delight given, but there's no question in my mind we came up positive this year. Certainly room for improvement in the future, of course, but this year ranks right at the top of my favorite April Fools.
Activity
Since the goal of the prank was to be a giant distraction, I wondered how much the design would affect activity on Stack Overflow:
era posts total_score avg_score questions answers
------ ----- ----------- --------- --------- -------
normal 31709 5612 0.176984 15669 16008
gap 85472 16029 0.187535 41657 43701
fools 32176 4459 0.138581 15642 16502
fools is the time from 2019-03-31 10:00:00 to 2019-04-02 10:00:00 UTC, normal is the same timeframe a week earlier and gap is the time between. (Once SEDE is updated to include this week's data, you can run the same query yourself.) There was a small increase posts due entirely to a 3% increase in answers. Conversely, the scores on those posts was down 20%. This was largely due to a 9% decrease in upvotes:
era votes upvote downvote
------ ------ ------ --------
normal 130304 92636 13522
gap 347243 249957 34223
fools 121414 84352 13358
I think the other reason scores were down on new posts is that a higher percentage of the votes went to older posts than normal. Spot checking a few other large sites, I saw generally similar results. The design seemed to have distracted people from exercising their vote on new content.
The story is radically different on Meta Stack Exchange:
era posts total_score avg_score questions answers
------ ----- ----------- --------- --------- -------
normal 103 -232 -2.252427 81 20
gap 247 -605 -2.449392 180 65
fools 240 2205 9.187500 100 138
era votes upvote downvote
------ ----- ------ --------
normal 1387 607 618
gap 3292 1375 1477
fools 4935 3493 1100
The event brought many more questions, answers (including on my question) and votes to meta. I don't think this is unique to this event, however. People just like to talk about whatever event is going on. Thank you to all the users who helped me handle the extra traffic on meta.
Revelation
While we were running a April Fools event that took the site back to the 90s, CommitStrip published Back to how it should be:
Now we don't think we ought to take the site back to the 90s, but one of the most common requests was to make the new theme permanent. If you are in that camp, I recommend taking a look at a stone arachnid's 2019 April Fool's Day Retro Theme for Tampermonkey. It really took me back to last Monday and beyond. If, like Catija, you just want the mouse trails, there's a script for that too.
We spent the better part of the previous year making changes to site designs. So far users have seen very little benefit since we took a short break from working on site designs. (We're finally gearing up again starting with Anime & Manga.) Now one of the factors for choosing this particular prank was to show the value of having a standard. With the exception of this bug, the changed theme worked the same everywhere: a crime against usability.
Thanks to everyone who enjoyed the 90s theme. While we do sell products and services, the true engine of our company is you, the community. Events like this are disruptive, but for a good purpose. Double thanks to those who endured the event. Hopefully it's been a relief to go back to normal. Fortunately, it's hard to imagine something we could do next year that would be more disruptive.
Addendum
One of the benefits of doing this sort of recap is that it brings things out in the open that might not have come out otherwise. A colleague from the team that handles support emails (via the contact form) told me their experience was a bit different than mine. Most years support tickets on April Fools Day tend to be playful jokes riffing off of whatever we did for the event. This year was different. Our support team came back from the weekend with double the tickets they normally get.
Some people thought the site was hacked and were trying to report it. Others wanted to make sure their information was safe. Some people told us the prank was unprofessional and that they were leaving the site. The negativity in the helpdesk was so intense the team wished we could just turn off the joke early.
Indeed, the statistics from Stack Overflow alone show this was a much busier stretch of time than the previous week:
era user_contacts moderator_deleted_users user_deletion_requests
------ ------------- ----------------------- ----------------------
normal 21 72 9
gap 49 113 24
fools 70 39 7
Deletions have gone down, which indicates users didn't make the ultimate protest, but we may very well see users not coming back to the sites in the weeks to come.
I'd like to thank our hardworking support team for shouldering the extra load. In my first few years of working as a community manager, I handled support emails one day a week and it is a thankless job. When people complain about the crap the see posted on the sites, I sometimes wish we could show some of the stuff we get via the contact form. Unfortunately the impact of this prank on their job just didn't enter my head until this morning. It's especially troublesome since I designed a feature that could have made a difference. We need to account for potential support spikes in future events.
This post is already so long, I might as well post some more graphs from Stack Overflow site analytics:
I use Stack Overflow because it has far more data, which makes unusual events more obvious. Posts and votes just followed the normal weekly pattern, but traffic spiked higher than normal. It wasn't just new people; all sorts of visits increased. My guess is the vast majority of extra visits were simply gawkers and not going to be productive members of the community.
