Museum Statistics - 2021 Edition

Who doesn't love a good graph?

I wanted to do an article last this last year, but between work and ZZT's 30th anniversary meaning a whole lot of streaming, I never had the chance. Hopefully this can be the start of a new tradition. This article will use the power of Microsoft Excel LibreOffice Calc to do a little number crunching and get a feel for what's been added and accessed on the Museum both in this past year as well as overall.

A few things to keep in mind, is that while I've done my best to make sure this data is accurate, there are of course plenty of instances where the data being worked with is inaccurate to begin with. Dates can be wrong, "unknowns" may actually be "partially knowns", and the kind of person that's got some interest in this site is likely the kind of person blocking analytic services in the interests of their own privacy (myself included). Much of this data is the sort of thing that is "accurate-enough". It serves as a good approximation, but do take any data presented here with a grain of salt.

• Uploads by Release Year
• Uploads by Platform
• Whose Files Were Uploaded?
• Releases Per Year
• Quality!
• Most Popular Pages
• Most Viewed Files
• Most Played Files
• A Year In The Queue
• File Activity

Uploads by Release Year

Let's begin by looking at files which were uploaded to the Museum in 2021. These files cover a range of thirty years for possible release dates. When do we suspect this year's batch of uploads were originally released?

Year	Releases
Unk.	81
1991	3
1992	8
1993	16
1994	21
1995	38
1996	51
1997	49
1998	18
1999	13
2000	6
2001	12
2002	8
2003	7
2004	9
2005	5
2006	2
2007	4
2008	1
2009	2
2010	1
2011	0
2012	4
2013	3
2014	2
2015	4
2016	1
2017	0
2018	1
2019	1
2020	1
2021	68
2022	1

First, the lone 2022 upload is just because I was working with a snapshot of the database taken on January 2nd, 2022 and forgot to cut that out of my data set. Whoops.

The trends are pretty clear here. A significant portion of 2021 uploads are indeed newly created titles. This makes sense, as they're not going to be uploaded anytime earlier of course!

The largest year still remains "Unknown". These are for worlds whose specific release dates can't be accurately determined. Currently, the Museum only supports specific dates, so if a game says 1995 on the title screen, but the file modification date says 2017, that has to get filed under unknown. This is something I hope to improve on this year!

Unknowns come from "tainted" files like ones that were re-saved even if no actual changes were made to a world. Incorrect dates may be used for timestamps to begin with, resulting in situations like this original version of Kudzu whose files are dated 1988, three years before ZZT's release! Some worlds were re-released later with the advent of tools to lock them. Other times, dates may seem reasonable, say a game with a timestamp from 1996 whose sequel is then dated as being from 1994. Needless to say, accurate dates aren't just a problem for archaeologists. Super, or otherwise.

So "Unknown" and 2021 have clear reasons to stand out, but how's the rest of that curve looking?

From past to present, it ramps up before quickly ramping back down. This makes sense as ZZT would have a narrower install base in early years, and pre-Internet files are significantly harder to come by. Things peak in the mid-90s during the AOL era where there's a sizable number of users, but archives basically amount to what gets posted on AOL. Eventually this archive goes down which requires these worlds to be rediscovered. Donations of backups from this era tend to be the most fruitful. It's also an era where a lot of what files are recovered are hardly lost classics. These files are of course still important and tell a story, but if you look at the various Publication Packs, you'll see a lot of very short, yellow bordered games that are more likely to be people experimenting with the editor rather than sitting down to create a complete experience for the masses in this era.

Even though the most prolific years for ZZT are in the late 90s and early 2000s, the uploads for that time period are also when things drop off significantly. By this time period, the ZZT community has generally (though not exclusively) become centered around "The ZZT Archive", which would eventually spin-off over a staff schism into z2. While alternative archives like ZeuxWorld also emerged in the years after, and of course there have always been those outside of the community creating with it in small numbers, these years have so little to recover because this is the era when ZZT worlds would have been least likely to be lost. Files recovered from this era are frequently alternate revisions of existing worlds.

Move ahead to the dry period beyond 2004ish, and at this point you have a ZZT community that's winding down. The number of already known releases drops significantly around this time. Compare 150+ files from 2003, to barely more than 50 in 2005. The year 2011 meanwhile still has zero files on the Museum. In this period, files being found are ZZT community expats or those who never interacted with it in the first place. The need for DOSBox and an inactive community on the main archive of z2 mean that these worlds are often quite novel creations that were completely unknown.

In 2015 the Museum opens and as archiving lost past works begins in earnest in the next few years it means that new releases are very focused on being uploaded directly to the Museum by the authors themselves. Still, "outsider" experiments crop up now and again showing that you can never be confident that any year in ZZT's history is in any way "complete".

Also, on December 31st there was some excitement to quickly upload some extra files to reach 69 for the year. This was RUINED when WiL posted a 2022 bug fix of Weave that was in the queue, replacing a 2021 file with one from 2022. Well, at least we have a goal for 2022.

Uploads by Platform

What platforms are these for?

Platform	Quantity
ZZT	401
Super ZZT	12
ZIG	4
Programs	24
Roms	2

Wouldn't you know it, the Museum of ZZT mostly had ZZT worlds uploaded last year.

2021 introduced us to Campaign Against SZZT, a world of unknown origin the espouses why ZZT is the superior game creation system of choice. Tim's wish was granted and this year gave us just one Super ZZT world for every 33 ZZT worlds, a crushing defeat for ZZT's less popular little brother.

However, if you look at Super ZZT worlds published this year, there are two uploads of Mutant Citadel, a parody conversion of Timcomput's Campaign Against SZZT done in Super ZZT, a new author's experimentation with a Puzzle Bobble engine, and the last release of the year (because we can't have nice things,) Pilk MaZE MAZE IN THE MAZE MAZE MAZE MAZE.... Nearly half of Super ZZT games published on the Museum in 2021 were made that same year!

Still, even with a dozen uploads to its name, by Super ZZT standards, this year was a blowout for the engine. Only two years themselves have had more Super ZZT worlds created, with seven in 1997, and ten in 1992, the first full year of the engine's release. (Yes, I'm realizing that this is comparing worlds created in a given year against worlds created in any year that were uploaded in this previous one. I... didn't think this through.) Asie's recent release of Super ClassicZoo also gives the platform more promise than it's seen in some time. I am expecting 2022 to be a decent year for Super ZZT, but with WiL's Weave ZZT also offering up a cool new toy to play with, only time will tell.

Something similar happened with the ZZT clone ZIG as well. Four files published thanks to a recovery of some very early releases of the ZZT clone that could. ZIG has a special spot on the Museum owing to its special status on z2 where worlds for it were officially welcome, hence ZIG having its own category rather than being lumped in with every other ZZT clone that never took off. There was also the first new ZIG release since 2002 thanks once again to irony with another parody conversion of Timcomput's manifesto.

ZZT clones which flopped make up plenty of this year's programs, though a few programs were created for the year as well. For the most part though, 2021's programs are ones still in active development, best found at their own release points rather than on the Museum.

Which is also where the roms come into play! Asie's work with trying to get ZZT running on older consoles and a GBA game jam resulted in Agent Orange creating 587 Squadron, intended to be played on the GBA.

Whose Files Were Uploaded?

Author	Quantity
Aetsch	4
Agent Orange	12
AKNeutron	4
Alexis Janson	4
Andrew Wagster	3
Aric McKeown	3
asie	5
Chronos30	5
Craig Boston	3
Dan Page	3
Darren Hewer	12
Dr. Dos	11
John Harger	3
KKairos	3
Knightt	4
kristomu	3
Lancer X	3
Mark McIntyre	5
Matt Williams	3
Nanobot	4
RT-55J	4
The Green Herring	4
TheDaveC	3
TimComput	4
Various	3
verasev	3
WiL	9
Zinfandel	3
ZZTrash	3

Hmmm, this is a tough one to make a chart for. I wanted to get an idea which authors stood out the most, but the reality is that a huge number of authors only had one or two uploads rendering the graph unintelligible at smaller sizes. This chart only shows authors with three or more releases, and also cuts out "unknown" and "various" from the list.

I was curious who had the most files uploaded this past year. For the bulk of authors, it's just one release, and even with this filter the counts are pretty low in general. Almost all the authors with noticeably taller counts are current day authors. When you look at who's had content recovered from the past, there's no real standout. Of course, except for Darren Hewer who actually ties the most releases for the year thanks to files obtained from a donated 5.25" ZZT backup disk from his childhood!

Dubious Counts Ahead - Releases Per Year

There's always a question of just how many ZZT games there are out there. When the Museum launched it hosted something like 2300 files including ZZT worlds, utilities, and anything else. Right now (Jan. 8th, 2022) that number is 3355. To think, I assumed the Museum was reasonably "complete" in what it offered versus what could ever be recovered or would be created in the future!

The Mass Downloads page kind of gives us the data we're after to a certain point. These compilations are generated nightly and divided by year. This makes sense for the number of worlds released in a year up to a point where things really dry out and suddenly instead of being by year, the 2010s are instead a single download for the entire decade.

This is also the case for the 2020s, although it genuinely might not be the proper approach for too long. The 2020s mass zipfile already has more files in just two full years than the 2010s mass zipfile! Let's actually break those years up and get better picture of ZZT's releases as we currently know them.

These totals are as simple as filtering by release year on the Museum and seeing how many entries get returned from the query. These are very dubious. Many worlds have nearly identical alternate versions, wrong data, may be in compilations where they're completely identical, bugfix uploads before the Museum supported editing unpublished uploads, contests which bundle many worlds that could have been individual uploads, and just overall are really really really bad at being anything more than a shaky measurement that's much faster to calculate than going through some 3000 zip files to properly handle duplicates, revisions, GBA ports... you get the idea hopefully. Oh, and not to mention the "Unknown" release dates that obviously mean every year has a lower count than it should.

And for good measure, since they have their own entries on the Mass Downloads, numbers for Super ZZT, Zig, Utilities, ZZM audio, and Featured Worlds zips as well.

Year	Quantity
Unknown	137
1991	24
1992	46
1993	88
1994	101
1995	159
1996	223
1997	233
1998	193
1999	320
2000	332
2001	404
2002	221
2003	155
2004	82
2005	51
2006	33
2007	86
2008	27
2009	15
2010	9
2011	ZERO
2012	18
2013	8
2014	7
2015	7
2016	7
2017	3
2018	12
2019	20
2020	50
2021	59
Super ZZT	54
ZIG	18
Utilities	133
ZZM Audio	26
Featured Worlds	126

Quality!

Year	Quantity
1991	2.875
1992	3.894
1993	2.00
1994	3.575
1995	3.031
1996	3.155
1997	3.134
1998	3.802
1999	3.5
2000	2.963
2001	2.827
2002	2.749
2003	2.289
2004	3.139
2005	2.691
2006	3.043
2007	3.484
2008	4.367
2009	3.725
2010	N/A
2011	N/A
2012	3.4
2013	4.5
2014	5
2015	5
2016	5
2017	N/A
2018	3.667
2019	4.556
2020	4.382
2021	4.417

Okay, this one is 100% a joke. What's a "good year" for ZZT games? Well, if we take the review scores, divvy them up into the release year of the file being reviewed, and then average that out for each year we can obviously clearly objectively definitively measure quality. It's science.

Really I just wanted to see how wild these numbers are. In addition to the numerous reasons why anybody could tell you this is a flawed chart, the reviews on the Museum mostly come from z2 imports which are from the early 2000s. A massive number of files on the Museum have zero reviews, and I would not trust this chart even if every file had a dozen. I was mostly curious if there would be a dip in numbers for the early years, but since many of those files have no reviews and those that would are more likely to be fondly remembered I guess this graph makes "sense".

Sorry to anybody who released anything in 1993. Nothing good must have been made then.

The good news though, is that only 1993 and 2003 are under 2.5. This graph confirms that on average, the average ZZT game is above average.

In all seriousness though, z2 reviews mandated a numeric score while the Museum defaults to not providing a numeric rating. I do feel like more recent years do have a better hit:miss ratio in terms of quality which is actually reflected on this graph, but uh try not to read anything into that other than "games made by people in their 30s tend to be better made than games made by people in their teens". Also note that while 2011 has no ZZT worlds to review, there are worlds from 2010 and 2017, and not one of them has been reviewed. Think of the impact you could have on this graph if you rated a world from those years.

Most Popular Pages

I was reluctant to share things like this as the data isn't something calculated from information stored in the Museum's database, but acquired via Google Analytics. Many folks, (myself included) will be running script/ad/tracking blockers and thus these are not numbers I would actually trust. I would wager as well that ZZT being a subject where users are more likely than average to not be counted in these numbers. I very much rarely look at these stats and have no interest in analyzing every little bit of data to squeeze more "engagement". I just want to count some numbers and get an idea of devices and screen resolutions.

But when I asked on Discord what kind of stats people wanted to see, most suggestions would be based on these numbers so :shrug:.

Absolutely take these numbers with a grain of salt. For one thing, Django changed its default method for trailing slashes at the end of URLs which impacted the Museum at some point this year, and so "/uploaded" and "/uploaded/" are counted as two distinct pages in the stats. I wound up combining them into one when the alternate version was also up there on the list to avoid duplication.

The top page shouldn't surprise anybody, but the second on the list was one I didn't expect. The "We Are Still Out Here" article was written in September 2020, specifically to get people up to speed who were part of ZZT communities in the past, but hadn't been to the Museum before. Most notably, this article is linked on the front page of z2, indicating that a good portion of former ZZTers are checking the old site and finding their way here.

Other popular pages are sensible enough, the upload queue, pages to download ZZT, pages to explain what ZZT is, new releases. Somehow ZZT clones makes the list? This is troubling and I will definitely need to update it in the future as it's a legacy article imported from z2 and cleaned up, but it's sorely lacking in any information on any but the most successful clones.

There are also both articles which feature scanned documents donated by Beth Daggert, covering ZZT newsletters and internal Epic documents. These are reasonable to have on the list as well as they appeal to non-ZZT fans and historians interested in the shareware era of PC gaming for MS-DOS.

And then there's The Green Herring's Cyber Purge which, while a great game stands alone as the only ZZT world in that list. The next ZZT world that would be on the list is ZZT v3.2 itself!

Most Viewed Files

Cyber Purge by definition has to top this list. I'm very curious why though! What is that world doing that others aren't

I asked The Green Herring if they did any kind of "marketing" and the answer was "Not really :(". Then I told them it was the most viewed ZZT world for the year.

Cyber Purge was released in February, obviously giving it a big advantage in time for people to see it over releases from later in the year. Looking into landing pages though it was down in the late 30s, meaning most views weren't coming from direct links. Here's what's likely the real secret though: It was visible on the front page for almost an entire month.

The date of publication was Feb. 19th, and I do specifically remember it getting a top spot to stay visible for as long as possible. The next 2021 releases that would have knocked it off the "always visible" spot were Maze Maze in the Maze and The Hall of the Kunger Binb, published March. 16th. This was also during the time I was still working a retail job so publishing was more infrequent. It's unlikely for a title to get that much time on the front page again anytime soon. (And again, it also helps that it had a very cool title screen, was a blast to play, got livestreamed, and I know I tweeted a video of the game's intro as well.)

ZZT itself takes the next spot, which is much more reasonable. Anybody that's interested in playing ZZT games anywhere other than in a browser will need to download that one. Super ZZT makes an appearance farther down, for the same reason.

The community project Town of ZZT Remix also has a considerable bump in numbers. This one did have the benefit of being retweeted by Tim Sweeney himself, which surely led to a few eyes on it!

Most of the rest of the list are various other big releases for 2021. Challenge of the Toulouse Hacking Convention is interesting in that it's likely something that brought in visitors that otherwise may not have shown up.

Sixteen Easy Pieces and Star Wench are the unusual inclusions on the list. The former being a release from 2005 that was archived when it was new. While I would still argue for it being ZZT's greatest puzzle game, I would not expect it to make this list more than fifteen years after it release. It is a featured world though, meaning it shows up on the front page with some regularity, but I am curious if the title screen alone is enough to explain its popularity.

Most Played Files

Meanwhile on for those who aren't looking at games in their browser, but actually playing them, we get a rather different set. Topping the charts is CraNKGod's Little Square Things, a puzzle game from 2001 with a (free!) modern non-ZZT version as well. Oddly none of the sites, (Itch, Steam, a personal page) mention ZZT so I am unsure what the deal is with this one. Maybe somebody just really liked the game.

ZZT itself is high up there, offering folks their first trip to Town or any of the other original worlds.

Another game by Anna Anthropy, Orc Killer is high up on the charts and I wonder if this is from specifically linking the play page on Twitter at one point and saying it's worth spending the ~5 minutes to complete. (It is! I love Orc Killer.) I suppose people are more likely to click if there's a human post saying "hey you should check this out" vs the automatic links to bot posted WoZZT tweets.

My Ludum Dare entry, Doug Tudeap takes a spot owing to being linked from the jam's page. I believe I didn't set up the browser playable version on Itch until a little later, so early players were asked to either download and set up ZZT or just play on the Museum. You can guess what they picked.

There's Sixteen Easy Pieces again. Now I'm wondering if the lack of an ability to get browser saves out of the browser is contributing to the numbers on LST and this one, both games in which I would very much not want to have to start over.

Yinglet Maker is doing quite well given how late into the year it was released! The Museum version was passed around for a day or two before it went up on Itch which would split up its play count. Kobold Smoocher also made the rounds for furries with a degree of separation from ZZT. Make your games star anthropomorphic characters. It's free marketing!! Thinking about it I'm a buffoon for making Doug Tudeap a dwarf and not a kobold. I am retconning this immediately.

I can't recall specifically promoting BUYSO's whose short length and charming premise lends itself well to playing in the browser, but it did get promoted in The Best of ZZT Part 2 - Modern Treasures, and is one with a lot of games with Itch versions plugged instead and encouragement to go there rather than the Museum.

...Actually that list includes Town Remix, Kunger Binb, When There Is No More Snow, and Star Wench from the previous chart. That would explain quite a bit actually.

Lastly though, at the end of the list Bomb Squad!, which is a true enigma. I have not streamed it, written about it, seen it discussed in modern times, promoted it, or have any reason for it to have an edge over any other game hosted on the Museum. If I remember correctly, it was a source of minor controversy for its content when it was new during a period with some backlash over offensive ZZT games like Teen Priest 2. I seem to recall Bomb Squad! often being the next game brought up when listing offensive games, but playing it for a minute, it seems more over the top and cartoonish so who knows. Anyway, I do not know why it would have a handful of plays to be put on the top of the list of some 3000 entries of URLs that start with /play/.

Oh, and the sensible stats to follow up here would be "most downloaded", but those are just direct links to zipfiles for the vast, vast majority of files hosted on the Museum and thus do not have analytics. Sorry!

A Year In The Queue

This was a fun suggestion if a rough one to calculate. Day by day, how many items were in the upload queue? This involves trusting me to do some bad math, and I do not trust my math 100% still. It requires taking the set of uploads that existed up to the current day and counting the ones with a publish date with a future date to the one being checked against. Checking against scrapes of the site on the Internet Archive lets me at least see a match in most cases, but the exact times of uploads/publications throws this off.

This definitely seems inaccurate at the end of the year when publishing and uploading was happening frequently, but I Trust the overall trends here. You can see asie's regular streams of unpreserved worlds just slamming the queue in the first half of the year, combined with a retail job making it very difficult to keep up. I quit that job at the end of April and that's where the queue starts shrinking finally, with a backlog that takes months to conquer before reaching a fairly steady rhythm of the rule of thumb for 8-uploads resulting in the up and down pattern. At least until Oktrollberfest and end of year uploads happen, causing spikes necessitating some larger/more frequent publication to settle things down.

I think as I'm writing this in early January, I may throw together a nightly script to just record these values rather than trying to calculate them again next year.

File Activity

Year	Views	Plays
2021	22,744	4,816
2020	14,680	3,061
2019	13,163	2,120
2018	6,488	1,030
2017*	1,172	219

* Analytics begin in August of 2017 so these numbers represent less than half a year

Let's end it with a rather pleasant chart showing how the Museum's growth over the years, which was particularly explosive in 2021! I am quite happy to see that site activity for the two biggest functions continues to rise with each passing year. Despite more than 30 years since its initial release, ZZT continues to capture an audience for itself, both in players and creators alike. Today it is freer than ever from its ties to long outdated hardware and remains a tool of creation with a unique blend of surface level simplicity and underlying complexity unlike any other since.

May the trends continue in 2022!

One For The Road...

This is the life I live.