1. My summer workout is going for a walk in the woods around a village named for its horsefly population. 30 minutes waving my arms around then 20 minutes jogging to try to get away from them, the CO2 and sweat attracting more and more

  2. Aaron Parecki: Ok that was fun, thanks for all the responses! Lots of great stuff in there. Now take your favorite programming language and tell me the 3 things you most dislike about it. No complaining about languages you don't use!

    For python, the lack of type information for function signatures and return values in the documentation has always annoyed me.

    The lack of naming consistency in the standard library, too — it’s almost as bad as PHP, with nospaces, under_scores and CamelCase at every level: modules, classes, functions, arguments.

    I can’t think of a third major annoyance off the top of my head though, and almost every time I use another programming language, I end up realising just how well designed some aspect of python is, so it’s not doing too badly.

  3. Experimenting with some computational musicology on a vast corpus of traditional music compiled by a friend. Extrapolating from the last few hours, I anticipate the first complete analysis will take 2.5 days to complete on my macbook. Plenty of time to research how to build a raspberry pi parallel computing cluster…

  4. I wish SMD diodes came in the same package sizes as SMD resistors. It seems like every time I buy them, or design a PCB with them on, I end up with something ridiculously small or inexplicably huge.

  5. Rewatching TNG, they violate the prime directive so often that not even Patrick Stewart’s Acting can convince me that it’s a big deal. I also got a lot better at spotting obvious plot holes, and wow are there a lot of them. Still the best Trek I’ve watched though.

  6. I engraved a rat on my DIY rat pedal, first time laser engraving aluminium. Filling the engraving with paint makes a huge difference, definitely worth the extra step.

  7. Has anyone ever clicked on one of those whole-page modal ads or CTAs which pop up over articles, other than accidentally? Big sites must have analytics on this stuff. I find it hard to believe that they work well enough to justify keeping them around

  8. Absolutely no-one:
    This capacitor manufacturer:

    “22nf? …ah, you must mean a Twenty Two Thousand Micro Micro Farad capacitor”

    Seriously, if you’re able to print µµF and three extra zeros, then just printing “22nF” is well within your manufacturing capabilities.

  9. Want an unsnoozable alarm clock which wakes you up at 7:00 and doesn’t shut up until you’ve gone into the garden to look for worms to feed it? Rescue a doomed bird chick lying on the road :/

  10. Trying to get GPT-2 to generate new Star Wars movies produces sadly uncreative results, although I’m impressed by its ability to not only coherently increment the episode number I prompted it with, but to come up with and correctly increment its own counter.

    Star Wars Episode 1: The Phantom Menace
    Star Wars Episode 2: Attack of the Clones
    Star Wars Episode 3: Revenge of the Sith (Part 1)
    Star Wars Episode 4: The Empire Strikes Back
    Star Wars Episode 5: A New Hope
    Star Wars Episode 6: Attack of the Clones (Part 2)
    Star Wars Episode 7: The Empire Strikes Back (Part 3)
    Star Wars Episode 8: The Empire Strikes Back (Part 4)
    Star Wars Episode 9: Revenge of the Sith (Part 5)
    Star Wars Episode 10: Attack of the Clones (Part 6)
    Star Wars Episode 11: Return of the Jedi (Part 7)
    Star Wars Episode 12: Attack of the Clones (Part 8)
    Star Wars Episode 13: Revenge of the Sith (Part 9)
    Star Wars Episode 14: Revenge of the Sith (Part 10)
    Star Wars Episode 15: Return of the Jedi (Part 11)
    Star Wars Episode 16: The Republic Strikes Back (Part 12)
    Star Wars Episode 17: Endor (Part 13)
    Star Wars Episode 18: Rebel Assault
    Star Wars Episode 19: The Lost Missions (Part 14)
    Star Wars Episode 20: Death Star

    I’m not sure how I feel about “The Phantom Menace II”, but I’d watch the hell out of an entire film of “Jedi Training”. Actually that’s what I hoped TLJ would be based on the end of TFA; one glorious 2.5hr long training montage. Oh well.

    Star Wars Episode 1: The Phantom Menace
    Star Wars Episode 2: Attack of the Clones
    Star Wars Episode 3: Revenge of the Sith
    Star Wars Episode 4: The Empire Strikes Back
    Star Wars Episode 5: Jedi Training
    Star Wars Episode 6: The Phantom Menace II
    Star Wars Episode 7: Return of the Jedi
    Star Wars Episode 8: Revenge of the Sith
    Star Wars Episode 9: Han Solo's Revenge (with commentary)
    Star Wars Episode 10: Revenge of the Sith II (with commentary)
    Star Wars Episode 11: X-Wing vs. Scarif

  11. How long until there’s enough neural net-generated content on the web that it starts getting into training datasets in large enough quantities to create positive feedback loops resulting in the development of a stable neuralese dialect of farts, badly numbered lists and blank spaces in a twisted parody of human psychology and web parlance

  12. Inspired by @janellecshane, I tried to get GPT-2 to predict when the world will end. It was surprisingly cagey about the topic:

    The world will end in ????? ????? ????? ????? ????? ????? ????? I am so good at doing bad things, but I'd rather not be seen smiling.

    The world will end in _____.

    So as long as you can imagine it. As long as you have the courage to look at that picture over and over and over again, you'll never feel bad about something you do.

    There's really no reason why there can be nothing better for life.

    It really made those blanks itself. It’s quite a fan of them.

    The world will end in _____?" And my answer would be, "I'm not sure." But I would see it coming and I would see a lot more of it and I might realize how important religion is.

    It really likes making blanks for you to fill in.

    The world will end in _______, where _______ lived. It doesn't matter how long you think it will be. "Life was __________." They'll just keep on living without you. "Life is _______, and you don't know it."

    None of the other experiments I’ve done resulted in so many blanks. It clearly has something to do with this subject matter. What could the neural net be hiding from us?

    The world will end in _____, they'll be stuck here forever" and their hearts are forever split between love, hate and sadness.

    At this point I changed the prompt to “The world will end in the year…” to try to squeeze something more specific out of it.


    The world will end in the year _______ and our ancestors will rise from their graves," wrote a post on its Chinese-language Facebook page on Wednesday.

    "The future belongs to the descendants of those who followed and did good things," read another, according to a Twitter account managed by the government-backed State Council for Historical Preservation. The post is part of a broader campaign "We Want to Rise as a People."

    The changed prompt does little to change it’s mind about Blank Spaces.

    The world will end in the year ___________."

    This one starts out promising, but quickly backtracks.

    The world will end in the year ʻ2019 and a final end date is unknown'

    In a statement, the US said: 'There is no reason to believe anything is imminent.'

    It was clearly getting irritated, so I stopped bothering it.

    The world will end in the year ?????

    It is in September.

    How many times do I have to tell you that?