My last time [of 70.02] felt beatable, but to be perfectly honest, this one feels like everything fell into place perfectly, so I would be highly impressed. –The version of Jacob who wrote my last blog post, who is to become highly impressed
I got a 58.16 on Seven Wordles! The game adds three seconds for every wrong guess, so that equates to 17 seconds real time, or about two and a half seconds a Wordle.
I wasn't planning to do more Wordle speedrunning after my last blog post, but on May 8, friend of the blog Prismatic got a time of 68.51, beating my personal record of 70.02.
On May 10, I got a 68.06.
On May 12, he got a 63.60.
On May 18, I got a 63.01.
On May 21, he got a 60.38.
And finally, yesterday, June 5, I got a 58.16.
I don't think it's actually the world record. I discovered a Wordle speedrunning Discord. Fortunately (for my record), they don't really do Seven Wordles much—they're into Speedle especially. I haven't seen video proof of a better run than mine, but I believe a user named egg, who has the Speedle record, said they got a 58.01.
I know that's only a 0.15-second gap, but this has consumed a lot of my spare time, so I really do intend to stop at this point. If Prismatic beats the record (which I think is likely if he keeps trying), I will not challenge it.
If Prismatic beats the record (which I think is likely if he keeps trying), I will not challenge it.
Anyway, finding this community was just a reminder to me that I don't know everyone on the internet, and there might be a better Wordle time out there.
[EDIT: The day after I published this, Prismatic did in fact beat the record! And with a 55.81: one of the words was SPILT. Also, he came out on the wrong side of a 50-50 at the end; if the last word had been CHASM instead of CHAOS it would've been sub-52. Regardless, very impressive and lucky!]
(By the way, I did look into creating a speedrun.com category for seven wordles, but that site prefers more substantial games, stating that the game "should have a reasonable length such as 5 minutes or longer as a casual playthrough.")
Okay, with that out of the way. How did I do it?
I won't lie, typing speed is a factor, as is luck. But there's also some strategy.
I continued to use my SPILT/CRANE strategy, a pair of words optimized for making as many words as possible uniquely determined. (You can get slightly better coverage if you use words that are not in the answer list, but I like having that slim chance of getting it in one.)
I used my program to tell me all the words that were uniquely determined after only SPILT, and tried to memorize them, with reasonably success. However, I was terrible at putting this memorization into practice. It felt like a for majority of times these words came up, I would type CRANE out of muscle memory only to realize it was TULIP, APTLY, SILLY, or one of the others, and be annoyed at my wasted guess.
Also, one time? My first word was actually SPILT. And my fingers, in their infinite wisdom, were so caught off guard that they decided to restart.
Anyway, in my ultimate speedrun, I got every word on three—no ones or twos to help me out. (A feat which, ironically, Prismatic speculated was implausible that day.) Last time, all of my threes but one were uniquely determined by then, but this time the opposite was true.
The chance of getting this lucky on all of them was 1 in 5*4*2*5*2*6*1=2400, which is pretty ridiculous.
LEECH could've been BELCH, EXCEL, FLECK, or WELCH.
PARER could've been PAPER, PAYER, or REPAY. (Don't know what possessed me to guess it.)
LATCH could've been OCTAL.
TACKY could've been BATCH, HATCH, MATCH, or WATCH.
SHEEP could've been SWEEP.
STOOD could've been SOOTH, SOOTY, SOUTH, STUDY, or STUFF.
Only LEASE was unique.
The chance of getting this lucky on all of them was 1 in 5*4*2*5*2*6*1=2400, which is pretty ridiculous. It is true that I've done a lot of resetting when the first few words don't go well, but it was still quite improbable.
Also, My Program Had a Minor Bug
The bug only affected a small number of cases, but was in a very basic function: the one that compares guesses to answers. It used to be that (to give an example) if SPILT was the word, and you guessed STILT, it would give you a yellow on the first T instead of a green on the second.
Essentially, it would calculate the colors of the letters in order, which is normally correct behavior—if the word was SPITE, the first T should be yellow but not the second. Unfortunately, a later green has priority over an earlier yellow. It never comes up when you guess words with no repeated letters, though, which I was mostly doing.
In Conclusion
I don't have too much else to say this time around! I think it's been the longest ever gap between this and my last blog post, and my most recent three blog posts are now all Wordle-related. I think that's going to change! I don't want to promise anything spectacular, but I have some topic ideas, and now that it's summer, I should have some more free time to actually write them.
Talk to you all then. Thanks for reading, as always, and the best of luck to anyone who wants to speedrun Wordle!
—Jacob
PS: Here's a custom wordle! Enjoy.