iiinova — where your digital self keeps evolving

Adam Harstad @adamharstad.bsky.social · 12h

Definitely want to load a metric buttload of compressed air into a multi-ton delivery platform and get it going 70+ miles per hour on public roadways, this is very practical technology. m.youtube.com/watch?v=9KDK...

m.youtube.com

What Happens When A Compressed Air Tank Explodes? | MythBusters

YouTube video by MythBusters

000 0

Adam Harstad @adamharstad.bsky.social · 02/10/2026

Elon Musk rather famously hates buses because… there’s other people on them. A clear and pitiable desire to secede from humanity. I wish they’d just get on with it.

020 0

Adam Harstad @adamharstad.bsky.social · 02/10/2026

Also hostile to concepts like common knowledge or shared reality. Infinitely personalized? You watch shows your coworkers watch so you can talk about them with your coworkers. You listen to music your friends listen to so you can go to concerts with your friends.

111 0

Adam Harstad @adamharstad.bsky.social · 29/09/2026

Gas prices (/cost of living) are obviously a big part of Trump’s current unpopularity. But so is e.g. deploying masked goons on US streets. And obsessing over the ballroom. It’s a big mélange and people hate all of it, every part has all mattered.

040 0

Adam Harstad @adamharstad.bsky.social · 29/09/2026

Generic ballot: * D +2.8 in September (fell as low as D +1.7 by early October) * D +4.4 in February * D +8.7 today. You can clearly see the big swing once ICE started ramping up. It clearly mattered. It changed minds. (I suspect it mattered more in districts that swung toward Trump in 2024.)

141 0

Adam Harstad @adamharstad.bsky.social · 29/09/2026

Here’s congressional approval from Gallup. 2nd-largest drop was from 3/26 to 4/26 (Iran war): net approval fell from -63 to -76. Largest drop was 9/25 to 10/25 (ICE surges): fell from -45 to -64. Texas wasn’t in play until recently because it began with a huge advantage, took a lot to chip away.

130 0

Adam Harstad @adamharstad.bsky.social · 29/09/2026

Trump’s net approval before Operation Midway Blitz: -12 (and stable) Before Renee Goode: -19 Before striking Iran: -22 Today: -29 There was no “tipping point”. Voters don’t only care about gas prices. Disapproval has steadily risen with each new policy. It’s cumulative.

Graph of poll averages from 9/7/25 showing 43% approval and 55% disapproval.

Graph of poll averages from 1/5/26 showing 39% approval and 58% disapproval.

Graph of poll averages from 2/23/26 showing 38% approval and 60% disapproval.

Graph of poll averages from 9/23/26 showing 35% approval and 64% disapproval.

150 0

Adam Harstad @adamharstad.bsky.social · 28/09/2026

As someone with ADHD, creating grocery lists is a massive burden that has often seemed insurmountable. LLMs aren’t the right hammer for this particular nail, though. (I use voice assistants with list integrations so I can automatically add stuff as soon as I think of it, even with my hands full.)

010 0

Adam Harstad @adamharstad.bsky.social · 26/09/2026

New “peekaboo” trend raises concern among infants about rash of parental disappearances.

020 0

Adam Harstad @adamharstad.bsky.social · 26/09/2026

Can I really remove my young son’s nose? I can at least act like I have.

130 0

Adam Harstad @adamharstad.bsky.social · 24/09/2026

Why? A slavish devotion to mediocre expectations? An aggressive lack of imagination? Quite kowtowing to convention, Hasan. Release your Early MVP candidates after Week 15. Be your own man.

070 0

Adam Harstad @adamharstad.bsky.social · 23/09/2026

I was thinking we could use them to replenish our war-depleted munitions stockpiles.

010 0

Adam Harstad @adamharstad.bsky.social · 23/09/2026

Alternately, we could increase electric shock glove funding 60,000 times over, that would also get us to our goal with with about $10b in walking-around money left over. Thanks for the ideas; this is a very hard problem to solve.

110 0

Adam Harstad @adamharstad.bsky.social · 23/09/2026

Brainstorm ideas to pay for the thing that costs negative one trillion dollars. ysph.yale.edu/news-article... But not DoorDash settlements or deferring pension contributions. We are serious people, so serious ideas only, please.

130 0

Adam Harstad @adamharstad.bsky.social · 21/09/2026

Same. www.footballguys.com/article/2026...

Screenshot of an article on LLMs. Highlighted text reads: “I ask him if he might similarly hire someone else to play his video games for him, and he is aghast. Sure, they might beat that boss in a third of the time and take less damage in the process but—regardless of what the narrative would claim—that boss does not need to be defeated. The world is not in danger. Time is not of the essence.”

170 0

Adam Harstad @adamharstad.bsky.social · 14/09/2026

All is well with me, Solomkon. Hope the same for you. Glad to see you still exist.

010 0

Adam Harstad @adamharstad.bsky.social · 14/09/2026

Probably not, but only one way to find out.

000 0

Reposted by Adam Harstad

Adam Harstad @adamharstad.bsky.social · 12/09/2026

Just published my favorite piece I’ve written since October 21, 2017. I am always tremendously appreciative that Footballguys lets me try shit like this (and endeavor to limit myself to a few per year so as not to abuse that trust). www.footballguys.com/article/2026...

footballguys.com

Dynasty, in Theory: Loving the Work

Thoughts on taking shortcuts, and on not.

3219 0

Adam Harstad @adamharstad.bsky.social · 12/09/2026

footballguys.com

Dynasty, in Theory: Loving the Work

Thoughts on taking shortcuts, and on not.

3219 0

Adam Harstad @adamharstad.bsky.social · 14/09/2026

As I always say, the nicest thing about a heuristics-based process is I absolutely don’t have to care *why* something is the case, only that it is. Lots of plausible explanations and I get to be indifferent about which is right because I know one (or, more likely, some combination) definitely is.

030 0

Adam Harstad @adamharstad.bsky.social · 14/09/2026

Not quite old yet, but working on it. And all is certainly well enough.

020 0

Adam Harstad @adamharstad.bsky.social · 14/09/2026

Alternately, could be e.g. measuring OL play, which should be largely uncorrelated with RB skill.

120 0

Adam Harstad @adamharstad.bsky.social · 14/09/2026

Not exceptions, examples. If a stat puts high- and low-quality players at both the top and bottom of the list, it’s evidence that the stat is measuring something other than player quality. (In this case, I’d say it’s more a “style of play” stat.)

240 0

Adam Harstad @adamharstad.bsky.social · 14/09/2026

Crucially, a forum-specific phenomenon rather than a sport-specific one. See also: Menswear Guy, the. It’s just that Sport was the largest forum-based subculture. All about those base rates.

000 0

Adam Harstad @adamharstad.bsky.social · 14/09/2026

Ironically, you’ve now said it twice and therefore have autoblocked yourself.

010 0

Adam Harstad @adamharstad.bsky.social · 14/09/2026

If your residuals look like the ones at the top, maybe consider the possibility that you’re not an analyst, you’re a hack.

120 0

Adam Harstad @adamharstad.bsky.social · 14/09/2026

Ideally you want something like this. Random spread, roughly equally above and below, looks like correlation is ~0. That’s probably a pretty good fit. (Well-behaved residuals don’t *guarantee* the fit is good, but it’s a damn good—and easy—place to start.)

110 0

Adam Harstad @adamharstad.bsky.social · 14/09/2026

If your residuals look like this, you’d probably be better-served with a different fit. Notice it’s high on the ends and low in the middle, like a parabola—you might e.g. be using a line of fit when the relationship is actually a curve (or a curve when you’d be better served with a line).

100 0

Adam Harstad @adamharstad.bsky.social · 14/09/2026

Teach time for those who never learned this shit: Whenever you drawing a best-fit line or curve, an easy check is to graph the distance between each point and the curve (the “residual”). Ideally: dots should be ~equally above and below, and there should be no visible pattern to the spread.

110 0

Adam Harstad @adamharstad.bsky.social · 14/09/2026

Freehanding that residuals graph (log-scaled, obvi). Yeah, looks fine to me.

Hand-drawn sketch of a graph with a bunch of dots clustered below zero on the left and a second cluster of dots off the page near the top right.

100 0

Adam Harstad @adamharstad.bsky.social · 14/09/2026

Lolllllllllllll graph your residuals, kids, lmfao

140 0

Adam Harstad @adamharstad.bsky.social · 14/09/2026

I think models have gotten better about tails in the last couple years, update your priors.

030 0

Adam Harstad @adamharstad.bsky.social · 14/09/2026

The only unjust penalties are spot fouls because they steal attribution for yardage produced from the proletariat (player stats) and assign it to the bourgeois (team stats).

051 0

Adam Harstad @adamharstad.bsky.social · 12/09/2026

Thank you. Obviously wasn’t entirely sure how (or if) this one would land, so I appreciate hearing this.

030 0

Adam Harstad @adamharstad.bsky.social · 04/09/2026

Desmond Howard’s record for single-season punt return yards (regular season + post) is probably the hardest-to-break record for a returner. But Lewis’ single-season combined yardage record is up there.

110 0

Adam Harstad @adamharstad.bsky.social · 04/09/2026

Turpin's probably closer than Dike is; he missed two games last year, over a full season he was pacing for 2118 yards. Still 15% (over 300 yards) shy of the mark, though.

120 0

Adam Harstad @adamharstad.bsky.social · 04/09/2026

Lewis' 2432 return yards from that year is more than 200 yards ahead of second place. Dike would have needed nearly 25% more yards last year to top it. It's going to take a lot for that record to fall.

120 0

Adam Harstad @adamharstad.bsky.social · 04/09/2026

The all-time record, though... I think that's pretty safe. In 2002, Michael ("Beer Man") Lewis became the only player in history to lead the league in punt and kickoff return yardage in the same season. Devin Hester is the only other player to lead in both in different seasons.

120 0

Adam Harstad @adamharstad.bsky.social · 04/09/2026

Chimere Dike had the 6th-most return yardage (1986) in NFL history last year. KeVontae Turpin ranked 19th, Myles Price ranked 32nd. I wouldn't be surprised if we get our 4th all-time 2,000-yard returner this year, especially with 17 games.

151 0

Adam Harstad @adamharstad.bsky.social · 01/09/2026

Like how scratching an itch feels so good that sometimes it was worth being itchy in the first place.

010 0

Adam Harstad @adamharstad.bsky.social · 31/08/2026

Always leave them wanting more.

020 0

Adam Harstad @adamharstad.bsky.social · 31/08/2026

ALSO WORTH POINTING OUT: 23 (soon to be 24) players have a career average of at least 11.3, and 7 (soon to be 8) of them got there with their returns from 2019-2025, so either 25% of the league has historically great punt returners right now, or we should... you know...

120 0

Adam Harstad @adamharstad.bsky.social · 31/08/2026

By way of comparison, if you remove all of Marcus Jones' 50+ yard returns his average falls to 10.4, which would rank 54th all-time. Not saying you should remove those long returns! They count! Just showing how boom/bust the position usually is. (Mims is the outlier, Jones is the norm.)

130 0