Just let it be clear that people are free to make suggestions for improvements down the line. We’ll start off in one way, but there is nothing really to say we can’t change things up to make it better suited to our needs later on. It’s always a bit of a trial-and-error I find with any discussion club! 
Also, I would suggests for you to also post the main link to the YouTube channel in the OP post, for ease of finding the channel. 
Good idea, I’ll put it in
I think it’d be fun to follow along with this! I’ve watched close to all Kurzgesagt videos in English already anyways so hopefully that will make it easier… Just quickly checked one and there’s a lot of science terms which I’m gonna struggle with but I’ll give it a shot.
I’ve always looked forward to the time I can learn Japanese through other subject areas like science, perhaps with school textbooks or something, so this is a step in that direction for sure!
My first check was the same feeling. 
I think the fact that we can automate extraction of vocab is a huge boon! I didn’t even think of that one.
And, I think, as always, the first video is going to be a bit painful as you settle into a new vocabulary, then things start to ease off. I really do wanna learn more science terminology in Japanese (if only so I can visit Osaka Science Museum | 大阪市立科学館 公式ホームページ in the future! )
Edit: Especially the Planetarium is something I’d love to visit! 
I just want the same kind of understanding ability I have in English, as in I want to be able to understand most topics, even if the words themselves are completely new to me.
Yeah, for sure. I totally get this feeling. I feel that with enough listening practice that is not an impossible goal at all. But the further outside of your current vocab knowledge you venture, the harder it gets to keep your feet on the ground so to speak.
We’ll get there, I’m sure, together! 
better than struggling alone for sure!
Polls are now closed, the results are pretty good, quite a few people said they would participate in some manner, and it seems most are for keeping everything in this thread, and only having a looser schedule.
As such, I’ll make a list of the videos that will be the candidates for the first watching in a second.
Before that though, the obligatory “When do you want to start” poll:
The poll is public
0 voters
And now a vote about which video we should watch. Like I said earlier, I will be including the three oldest and the three most recent videos in this:
0 voters
ダイソン球への道ー 究極の巨大構造体: Astrophysics. Dyson spheres are theoretical megastructures, that could power our future civilization, the video talks about the theory itself, the pitfalls and the potential ways forward
存在しないはずの（でも存在する）ウイルス: Biology (Immunology). The video talks about giant viruses, that were recently found
地球温暖化は防げる？: Global Politics. The Japanese title is a bit misleading, the original English title was “Can YOU stop climate change?”. Talks about how an individual can affect climate change, if there’s any benefit to being individually green, and so on.
ワクチンの副反応ーリスクの高さは？: Biology (Immunology). This was I believe originally a COVID related video. As the title suggests, the video talks about vaccines, their risks, their benefits, etc.
楽観的虚無主義: Philosophy. Talks about how to look at life optimistically, even if you feel like nothing you do maters in the grand scheme of things.
星々へと続く1,000kmのケーブルースカイフック: Astrophysics. The sky hook is a theoretical concept, that, if built could lower the cost on gettings things into space dramatically. Quite similar to the Dyson Sphere one.
As you can see, I included the runtime of each video. This includes the merch and sponsor plugs at the end and the intro sequence, therefore it is heavily skewed, when it comes to shorter videos. After the voting period, the actual raw length of content in the video will decide how long the club will be watching along the video. Specifics TBD.
I’ll close both polls in approximately 3 days. So about 2023-06-01T05:39:00Z
Interesting to see how the vote changes over time. Which options are the most voted for and which are the least voted for. For example “存在しないはずの（でも存在する）ウイルス” fell from one of the top places down to the last, but now last place has way fewer votes.
Just found this thread, so I can’t vote in the initial polls, but I’m definitely interested in joining! The proposed starter videos all look so interesting. 
Those were only for gauging interest early on, so no need to really. They’ll be replaced tomorrow most likely by an actual club description.
And the vote for the first video we’ll be watching is finished. Winner is “ダイソン球への道ー 究極の巨大構造体” (with a single vote), which is the very first video on the channel. Fitting.
We’ll be starting next week on Saturday, according to the usual book club timing.
Now for the fun bits. I was originally thinking about a schedule where for each 3 minutes of video we allocate a week. This would be nice, because most videos don’t go above 12 minutes, so most of them would end up being watched in 2-4 weeks. It should be a decent enough pace for those that can finish the video in one sitting to not have to wait for an eternity while also letting the beginners go at a comfortable pace. However, if someone has any objections, or other recommendations, feel free to mention them.
The video has a raw content length of 7:40 (7 seconds of intro from 1:15-1:22 and the sponsor segment starts at 7:47). This would put this video at 3 weeks, with about 2:30 each week.
In the meantime I’ll try to get a system going, that can rip and parse the subtitles.
Great stuff!
You’re the boss for how to run this club, however, I will watch the whole thing in one go since it’s just 12 minutes. XD Or in 2 sittings at the very most if it gets overwhelming.
It’s a matter of narrative flow really. The 3-minute-mark isn’t necessarily a natural place to stop. It might be midway through an explanation where you need both sides of that 3 minute mark to understand.
(And, just generally speaking, the thing with listening practice I find, is that you need to get dug in to latch onto the themes and narrative. If you break things up too much by constantly pausing, it just makes it harder to get what it’s all about. )
So, yeah, that’s why, I’ll be doing full views of the video, but multiple times to get the stuff I don’t get the first time round.
Edit: I gave it a test watch and it should be fine to do the whole thing. it will take longer to get every details of it however. ^^’ Starting off with having a look at the transcript for watch # 2.
Getting subtitles is as simple as ⋯, Show Transcript, then Copy+Paste.
I’ll do the same most likely, and I did mention that intermediate watchers will probably just watch the thing in its entirety. The 3 minute thing is there to give enough time for everyone to watch the video, even if they get overwhelmed after the watch. There won’t be like proper weekly breakdowns, only the start of the club and then 3 weeks later the end.
That’s both very manual and leads to a format that’s not as easy to parse. I’d much rather just use yt-dlp and rip the subtitles with that.
Well, spent a bit of time coding up a word extractor, and this is the result:
It probably has words missing, it most likely has incorrect readings, but it’s probably good enough and usable.
Now I’m even more interested in the video. Must be hell of a topic to reach from Dyson spheres to Touhou.
Asking out of interest: how did you create it?
I was thinking of making my own Mining flashcards by extracting the subtitles of videos sentence by sentence, and then passing them through ichi.moe (either through their API or running it locally via ichiran). Then with additional filtering of course. But open to suggestions how others made it.
That’s the translation source, it should prolly be ignored.
I mostly copied the python script ChristopherFritz wrote. The original took the output from mokuro, fed it to Juman++, which is a morphological analyzer, so it gives you the words in the sentence (more or less), and then that is used with just a hardcoded wiktionary list to give you the meanings.
My version differs from this by first using a youtube downloader to rip the subtitles from youtube, converts them to lrc format, because that’s relatively simple to parse and then feeds the individual lines into Juman++. Also, it’s in Ruby, because I don’t speak snek. But the end result is the same, a list of kanji-kana-timestamp triplets that then get pushed into the google sheet and paired up with their meanings.
After this personally I would just use a local copy of jmdict to look up the individual words. I wouldn’t recommend ichiran, only because you really don’t need all that information, and it’s not simple to set up, while jmdict has “”“bindings”“” for all sorts of languages.
Thanks for the very detailed explanation!
