The Dink Network

So, the D-Mod Download Statistics that I posted before were very tainted by bot downloads. As it turns out, there are quite a few bots that try to download every file they can find on the internet. I recently updated the robots.txt file to ask them to stop it (except the Internet Wayback Machine - they're cool).

Someone did try to use wget to download most of the files off of the site, but I didn't exclude it because I assume it was controlled by some sort of human being.

Much thanks for Tim for pointing out how weird it was that an old version of Malachi was downloaded a suspicious number of times.

So, I figured a new topic was in order. With good numbers representing people instead of machines playing our games.

D-Mods aren't quite as popular as they appeared before, but they're not too bad.

These download counts only cover downloads from Nov 17 2014 to Jan 15 2015

(I thought I'd try to make that more obvious for any pointy fish who might be reading this).

1. D-Mods by Rating



Here's every D-Mod downloaded more than 20 times in the past two months (that has at least one review), plus a few extra D-Mods with high ratings or Epic status. I also decided to change the axes on the chart; higher downloads appear further up the chart, and higher ratings appear along the right.

The biggest difference in this chart is the total download counts are much lower. Before, it looked like quite a few D-Mods had more than 100 downloads, but there's actually only 6 that break that barrier.

Attack of the Evil Wizard is still a bizarre outlier; I'll get into my suspicions later.

2. Dink Smallwood HD



With the previous charts, I just couldn't understand why Agatha, Furball, and Three Amulets were getting such high download counts; then I realized that it is possible to scroll the D-Mod list in Dink Smallwood HD by clicking-and-dragging the screen like it is a smartphone. Argh.

Dink Smallwood HD links to the following D-Mods:

Savebot Massacre: 147
Cloud Castle 2: 156 (old version)
Initiation v1.0.3: 125 (old version)
Lyna's Story: 163
Agatha Smallwood's Will: 82
Furball: 77
Three Amulets: 78(old version)

Of these, Seth is linking to several old versions. So the totals for the old versions are likely all from Dink Smallwood HD.

There is a very strong bias for the Dink Smallwood HD D-Mods, they're downloaded more than their rating would suggest. But, that makes the download counts for Lyna's Story a little surprising: it's the #2 rated D-Mod on the site, and linked from Dink Smallwood HD, but it's still 20% behind Pilgrim's Quest.

3. Epics



Epics don't seem to be downloaded more than their rating would otherwise suggest. Epics are generally rated higher than other D-Mods, but they don't seem to be necessarily downloaded more than comparably-rated D-Mods. There's a huge gap between Pilgrim's Quest and Historical Hero II.

Pilgrim's Quest: 201
Historical Hero II: 84 (also updated about a month ago)
Stone of Balance: 82
Cloud Castle 2: 73
As Good As Eternity: 69 (also on front page)
Prophecy of the Ancients: 55
FIAT: 43
Quest for Dorinthia Special Edition: 39
World of DinkC: 34
Historical Hero: 34
Quest for Dorinthia 2: 34
Scourger: 27
Quest for Dorinthia: 12 (the not-so-special edition)
Secret of Parizaya: 8

4. 1st Page



Here are the D-Mods that appear on the first page of D-Mod results. I think this explains why Attack of the Evil Wizard is downloaded so often; it's a nice simple title, even if it is rated 3.8. Agatha might also be getting a front-page bump.

Agatha Smallwood's Will: 82 (also linked from Dink Smallwood HD)
The Blacksmith's Trail: 77
As Good As Eternity: 69
Back from the Grave: 41
9 Gems of Life: 39
Broken Windows: 34
Attack of the Evil Wizard: 31
A Day in the Life of Dink Smallwood: 30
Adventures of Dink Smallwood Part 1: 27
Alex (new character): 25 (interestingly, just a developer D-mod)
A Fish Named Bloop: 22

5. Ratings >= 9.5



For D-Mods rated greater than or equal to 9.5, there's a pretty clear pattern. The biggest oddity is the One Screen D-Mod Compilation; most people must want to play something a bit more substantial.

Pilgrim's Quest: 201
Lyna's Story: 163 (also linked from Dink Smallwood HD)
Initiation v1.1.0: 106
The Blacksmith's Trail: 77 (also on front page)
Cloud Castle 2 1.01: 73
Power of Blood: 61
One Screen D-Mod Compilation: 26

6. D-Mods by Rating



This is similar to the first chart, but it orders the number of downloads (now on the x axis... sigh) against the ranking of the D-Mod as sorted by rating descending.

HH2 silver is downloaded more because it was recently updated, and Furball is linked from Dink Smallwood HD, and As Good As Eternity is on the first page. Everything else seems to line up fairly well.

6. Recent Releases



And now for the real download counts for recent releases. These weren't affected too much by the bots, but the numbers did decrease a little bit.

7. Data

D-Mod Download Totals (CSV) (Excel)
D-Mod Downloads by Day (CSV) (Excel)

I'm really starting to hate excel.

It seems screwy that COTPATD 2000 has 333 downloads. COTPATD 1998 has 20, for comparison.

It's also COTPATD 2000 v1.00, not the latest 1c.

The user agent strings look fine; there's a couple that are directly linked from google search results, but most look like normal requests. But, I did a reverse DNS lookup on several of the IP addresses, and they do look like bots.

Now this is more to what I expected even though it's only from November last year.

I decided to run AWStats on my server (the one hosting Dink Solutions) to see what it would pick up on. Most visitors are actually Chinese people who want to break in by running an SQL injection or something. For Oct '14 these were the most popular walkthroughs:

The most popular walkthrough is Cast Awakening 1 followed by Cloud Castle 2, then Stone of Balance. (The list on the main page isn't really accurate as Mystery Island has had very few views)
Malachi the Jerk got 70 views and "Dink and the 4 Towers got 48". Historical Hero 2 got 37 views in that same time period, although I haven't filtered bots out.

Lots of people are searching for a walkthrough for "Malachi the Jerk" and the search terms are hilarious at times:


dink smallwood mystery island walkthrough
dink smallwood walkthrough
dink smallwood walkthrough furball
what are some of the secrets or technics you can to get rid of scorpions
walkthrough for dmod malachi the jerk
is malachi a jerk or nah
dink smallwood cast awakening walkthrough
dink cheese


I'm looking into Piwik now because it's capable of analysing Apache logs and appears to do a much better job of it.

Dink Cheese.

Uh...

I love "is malachi a jerk or nah."

You know, it concerns me a bit that the scores have so much impact on how many people play these things because they're based on so few data points compared to other sites that have ratings. There are fewer points than ever nowadays as hardly anybody is writing reviews. I think it might be better if people could submit a simple rating.

I certainly would like to have a seperate rating system that didn't involve reviews. You'd have to have an account to set a rating, and that rating wouldn't be anonymous. As in, it would show up on your profile just like a review would. Currently I have set the fates of many a dmod and I do not wish to have that kind of power.

I'd also be happy -though less so than if a rating system was added- if the score section simply had a number next to it to signify how many reviews that score is out of.

As shown recently, we're already having problems with people agreeing on the given scores as it is, even with reviews that point out pros/cons. I can't see how it would work very fairly if people could just hand out the scores they wish without any basis whatsoever. The reviews act just as much as a means for the staff to judge whether the score is fair and given for the right reasons, as they are to give insight of the files, to their readers.

Well I mean, I said separate so the current review system wouldn't actually change. Instead there would just be another rating system.

I think the best way to respond to a score you disagree with would be to vote yourself. The reason scores have to be "justified" as it is (I've hardly seen this anywhere else on the Internet where people rate things) is because one score has so much influence on the overall score. Hopefully with more people voting, this would no longer be the case and the score would more accurately represent what the community thinks.

Yeah, I wouldn't trust a non-reviewed click to vote kinda thing. Even if it was separate, every DMOD would recieve low scores by trolls and scare away potential downloaders.

I agree with Skull on this one. A free-for-all scoring system doesn't work on Metacritic, why would it work on TDN? (Although I guess Metacritic's is techincally a review system, which isn't moderated at all/is moderated very little)

I don't necessarily see a problem with a file only having one review. Let's say Leprochaun writes a review giving a good dmod an 8.0, for example, and no one else writes another review in years? I think that's a sign that the score is pretty agreeable. If I see a score that I find too high/too low, I WILL write a review for that file. I think that works. There aren't many (if any) dmods here that currently have scores they clearly don't deserve.

Not having any reviews at all sucks, though. Write more reviews, people. With a simple scoring system, though, I think we'd just end with a lot of 9.9's and 0.0's, and as evidenced by Metacritic, that doesn't necessarily mold into a fair end score. With the current system, at least it's easy to influence the ratings by giving your own honest opinion - the same isn't true if you're fighting against a hundred out of spite/fake account 0.0's.

EDIT: Actually, you can just give a score on Metacritic, or give a score and write a review along it. First time I actually bothered to create an account and vote on anything in that cesspool... Obviously, I gave a 10, because anything else would just make my vote count for less. =)

The way to fix it would be to get Redink to change how Dmods are sorted by using an algorithm. I read this article years ago and immediately thought of this site.

Yep, but that algorithm only works on things-up/thumbs-down, not a score based system. I was looking into this a month ago, and have some tentative plans to change some things and encourage participation.

Come to think of it, there is a problem with votes and sorting by them. You see a thing you think deserves an 8, but it has a 4. Your 8 would move it very little. You think "that average is ridiculous, I'm going to have to give it a 10 to have as much of an impact as possible." People don't vote what they really mean when you're taking an average.

He more recently explained how to apply it to graded star ratings. It looks rather complex though.

I recon it would be very encouraging to find out how many times a dmod you made was downloaded. Maybe with those figures beside a download authors will get cracking and do more. But as it is all you get is the same people commenting on it usually without reading instructions if its different to the norm. My 2 cents worth gawd i love that emote heh