TASVideos was represented at Awesome Games Done Quick 2016 by dwangoAC, micro500, and others. TASs using a TASBot were shown at 2016-01-09 at around 20:00 UTC. More information is available in our discussion thread and the official website for the event.
Nach: This year, TASVideos had roughly a 2-hour block to show off TASs and TASing related technology. During this block, TASVideos helped raise ~$65,000 for the Prevent Cancer Foundation, contributing towards over $1 million for Awesome Games Done Quick 2016 as a whole.
There are plenty of other interesting runs on the official AGDQ 2016 schedule that may be of interest.
Nach: After this event, I reviewed logs from several IRC channels discussing this block, as well as forums and other discussion groups all over the Internet in order to gauge how people felt about our demonstration, and what impressions they were under when watching them.
Mario Kart 64 - Speedrun
The reactions to this segment were limited, but overall positive.
Humans Vs. TAS Competition
Unlike previous competitions, there were significantly fewer submissions generated for this one. A large sentiment among players were that the game was too open to effectively study and run within four hours. Overall reaction was lukewarm.
General viewers greatly disliked this segment. They felt the game was unknown, and the objectives unclear. They were uncertain as to what they were watching and had no way to gauge the progress of the different runners. Coupled with this, the game's name was not displayed on the stream, and it was not mentioned during the course of the run, preventing viewers from searching for additional information.
I could not find any discussions which had a great liking for the segment, and many were vocal regarding their displeasure.
The Legend of Zelda II - Speedrun
The reactions to this run were mostly surprise. Viewers were amazed to see this, and were somewhat unsure of what they were seeing. Overall extremely positive.
Brain Age - Drawings - Twitch
At the beginning of this run, viewers were very impressed. They enjoyed the initial jokes and other fun things the run did. As the run progressed, those watching it live in various official mediums had a lot of fun trying to predict what pictures would be. They enjoyed the fun poked at the viewer, and they loved playing along. Over Twitch, the drawings which were represented in twitch emoticons were especially popular.
Outside of official channels, many viewers felt the run dragged on a bit too long. But overall, the responses were positive, and most of those who got bored were still impressed by it.
A common sentiment for this run was that the game was hacked to accept any answer as correct. Some found our detailed resources for how this run worked, but the overwhelming majority did not.
Several viewers thought the answers via Twitch segments were simply staged, as the answers chosen were culled from Twitch from the moment that TASBot indicated Twitch was loading and was still discussing previous drawings, not from when TASBot actually started drawing answers to the question. This led to a disparity between what was shown, and what was being seen in the chat itself (such as knuckles instead of 7).
Super Mario Bros 3. - Speedrun and Total Control playaround - Demonstration
The viewers went absolutely mad regarding this run. The iconic levels with the new features drove the viewers into a frenzy. Many wanted to know where they can download this hack of the game that was played here. Aside from the general lack of understanding regarding this run, the reactions were overwhelmingly positive.
Many took Toadstool's hacked message at the end as confirmation of this being a hack, failing to realize that both the original and this version begins with an antithesis of the situation which is later negated. More than any other segment, this run needed clarification at the end that it indeed was not a hack.
Super Mario World - Total Control - Demonstration - Twitch
At this point in the block, outside of the US timezones, many viewers were getting tired and went to bed. Seeing that this was another Super Mario World demonstration caused many of the viewers to think this would be another repeat of previous years aiding their decision to go finally get some sleep. Therefore, overall response on this segment was rather limited.
For those that did watch, many were mildly impressed at what was done to this run. Many viewers also complained that Nintendo has been ripping them off for years by only making such features accessible recently in their games, while Nintendo had this built into Super Mario World the entire time.
We did an excellent job this year, and knocked the socks off of viewers. We also (re-)introduced TASs to a huge amount of people. The viewers were mostly entertained, and found the block as a whole exciting, and are salivating for what we will do next year. However, there's still plenty of room for improvement.
There were a large amount of viewers who thought what was done for these runs was simply unlocking existing features heretofore unknown in games, or else playing some kind of hacked version, probably more explanation was needed at the end of each run to clarify what viewers just saw. Many viewers also expressed sentiments that R.O.B. (TASBot) was some kind of artificial intelligence, and did not understand that it is simply a replay bot, or a bot which receives instructions online. Perhaps an overall explanation at the beginning and end of the block as to what TASBot is (and isn't) would help.
Several viewers, and even one vocal donator live during the block seemed to be unclear as to the existence of TASVideos. They seemed to be under the impression that TASs are something reserved for doing weird hacker things at Games Done Quick. Perhaps the site's URL should be on display in the corner of the stream during our block.
Nach: Many of our regulars were not visiting the site as usual surrounding the time of the block, as they were working on the speedrun competition, watching the block itself, or resting to prepare for it. Many were also worn out from the block and went to bed immediately thereafter (or towards the end of it). Despite this, the site had a huge increase in activity during and surrounding the block.
Towards the end of the block and shortly thereafter, the site received 82,726 hits, which is roughly 10 times what it sees at a similar time frame. During the surrounding hours, the site received 219,863 hits, which is roughly four times the amount of similar time frames. A typical day, has ~150,000 hits, whereas the day surrounding the event generated 646,529 hits. It was also a stroke of genius to display an unexplained QR code for viewers to figure out on their own at the end of the Brain Age run, as it led more people to the site in the hunt of a surprise of some kind. Contrasting to just displaying the site's URL on the screen, which would probably be ignored by nearly everyone, this worked out well.
DeHackEd and I (Nach) spent some time the week leading up to the event to ensure the server would be able to handle the load, and it did quite well. Despite the large increase in hits, from the server's perspective, the increase was minimal, and did not lead to any noticeable performance penalties in the slightest. This contrasts with previous years were the site was either down for a while or somewhat unresponsive and sluggish. With these changes now in place, we appear to be ready for handling significantly increased loads.
It seems we are still having a nice residual effect from the event as our activity per day is still quite high compared to what it was prior to the event.
- 2016-01-11 - 279,734 hits.
- 2016-01-12 - 267,053 hits.
- 2016-01-13 - 264,552 hits.
- 2016-01-14 - 264,753 hits.
- 2016-01-15 - 254,721 hits.
- 2016-01-16 - 233,603 hits.
- 2016-01-17 - 236,789 hits.
We've also seen forum registration increases somewhat above what it usually is.
- Ars Technica article on how Super Mario World was turned into Mario Maker, along with behind the scenes photos of what went into preparing some of the other segments.