I've appreciated watching a lot of the runs here, and I'm hoping to add some of my own in a new category I'm trying to develop. Right now, we have 'set seed', which gets hyper-specialized and a huge part of it is finding the right seed, so completion times are pretty short, and then we have random seed, where RNG and drops seem to dominate, and if you don't get a relatively flat biome like e.g. desert at the start you abandon the run. Also, as minecraft changes and updates, many of the WR runs are still on rather old versions or easy modes because it enables better times.

I'm trying to create a category that combines traditional speedrun optimization with current versions of minecraft. My idea is to have an algorithm generate a new seed for every quarter, and then to only allow runs within that quarter using that seed, on versions that were current in that quarter, and require hardcore mode. So, for example, my prototype algorithm has the following seeds generated:

THESE SEEDS ARE ALL OLD NOW: 20190101: -531774365177366475 20190401: -6160165691971503792 20190701: -1192446498326702188 20191001: 3418289440987301865 20200101: 9051420377272009586 20200401: -3306925433473001243 20200701: -6500560943761209387 20201001: -7206277944359803878

Because we're in the quarter between 20191001 and 20200101, the seed for the next couple months would be OLD ALGORITHM SEED, NO LONGER VALID: 3418289440987301865. Versions allowed would be 1.14.4, which was current at the start of the period and is at present, and any other versions that are released as current (not development) within the time period, so maybe 1.14.5, 1.15 or 1.15.1, depending on what happens.

What are your initial thoughts? Would you be interested in watching or running this kind of a thing?

Okey, I have to say that I am a Little bit confused with the seeds, so pls be more specific with that if is posible, is more confortable, like (example, no true lol lmao): December of 2019: -531774365177366475 January of 2020: -6160165691971503792 February of 2020: -1192446498326702188 etc, etc, etc.

In all case, WOW! I like the idea, would be so nice yes. With the idea that I understand, is basically like another "set seed" category, that all the people try to be the best in that seed, but with anothers diferents seed that are not too worth to a set seed, so I would like that variety in set seed, having a different seed in each month, to be variet.

Think in all the time that speedrunners use in so many games to get WR... I help you with an example: in "New Super Mario Bros Wii", in Any% category, look the WR of FadeVanity; MORE THAN 3000 ATTEMPTS! For a category that now the WR is of 24:29 ! WOW! Basically, I say this to understand that the people can be much time, but SO SO SO MUCH time with a game without changes, like in Minecraft set seed, that the world is always the same and you always do exactly the same.

So, yeah, that would be like a mini-competition: The first day, you give an aleatory seed,and the people have like 1/2/3/4 weeks / 1 month to investigate it and run it, and we look finally who get the best time. Think that, if you make now the seed, I can run it today, and submit 7 weeks later, so I think thats not the idea.

Only a recomendation: if a seed of the list is for, for example, for the january month of 2020, dont reveal yet, and share it in january of 2020.

Ps: the same seed have diferents worlds in diferents versions, so each 1 can only be for 1 version lol. Ps: apart of the current version, would be so nice seeds of older versions. I am player of 1.8 lmao.

@KilleDragon:

Thanks for your feedback!

I'll try to explain better for the seeds. You have the right idea, where it's a set seed but it's a seed selected by an algorithm to be random-like, and we generate a new seed every 3 months in a particular well defined way. Because the seed is chosen in this way, it could be good or bad, but we compete to do the best on that one. My thinking is that a new seed every month is too frequent, and doesn't give a good chance to have several different strategies evolve, or encourage people to build better tools to help planning the routes. So my current plan is a new seed every 3 months (sometimes people refer to these periods as 'Quarters', so a possible name for these seeds would be e.g. 2019Q4S for the 4th quarter seed in 2019, e.g. October through December 2019). The way to read the seeds from my initial post is:

20191001 = October through December 2019: 3418289440987301865 20200101 = January through March 2020: 9051420377272009586 20200401 = April through June 2020: -3306925433473001243 20200701 = July through September 2020: -6500560943761209387 20201001 = October through December 2020: -7206277944359803878

On your recommendation, it's possible for me to make the algorithm so that no one knows the seed until the quarter starts, by mixing in some kind of reliable external data, e.g. stock price close on the day before the quarter.

This makes the algorithm require a connection to the internet to pull the data, and it requires whatever site is serving that data to still be around, but it's an acceptable method used in e.g. geohashing (https://en.wikipedia.org/wiki/Geohashing)

On whether or not to allow older versions, I feel like the older versions are well-enough covered with the existing categories, and part of my goal in creating this approach is to incentivize speedruns on more recent and diverse minecraft content. Because the seed generates differently on older versions (at least 5 different generators, one each for <1.6, 1.7-1.8, 1.9-1.12, 1.13, 1.14), I think it would make the viewing experience for someone comparing runs worse.

I'm going to work on a version of the generator algorithm that won't allow knowing the next season ahead of time, and then I'll post the current seed here.

Okay, with the current algorithm I've factored in the final close from the S&P 500 into the seed for the given quarter, that way you can't know it ahead of time. The seed for this quarter, October through December 2019, is 7095016954182664219.

The algorithm is pbkdf2(sha256, 'YYYY-MM-DD-$$$.$$', salt='minecraft-quarter-seed', rounds=42)[0:8], interpreted as a signed little endian 64 bit integer. YYYY-MM-DD is the first day of the quarter, in this case, 2019-10-01, and $$$.$$ is the last close of the S&P 500 before the start of the quarter, in this case 2976.74. This yields

2019-10-01-2976.74: 7095016954182664219

My first attempt ended in a quick death:

I'm new to speed running minecraft so most of my actions aren't really well thought out.

The seed spawns on a tiny island near a few larger islands, but it seems like a trip through the nether is going to be necessary to get somewhere better quickly, most likely.

I'm doing some recon, it seems like the best of the strongholds to go for is near -892 ~ 2201. It has 2 eyes in the frame, and another two available in chests. but its generation is a bit cut up because it's near an abandoned mineshaft and a dungeon.

NICE! I like it lol lmao

@simplesymbols by the way: You figure out about the speedrun video time is LITERALLY 20:19? jaja XD

The speedrun have to be in hardcore mode, no? that seems cool, an important info

Yeah, hardcore mode. I'm doing research on the seed in creative mode, and I may practice in hard. I think I'm going to make another thread to capture runs on this season's seed.