Lightnews — Scholar-powered news

mark thompson @markposts.bsky.social · 24m

I find it kinda funny that there are companies incentivising token use, but it also seems that a full/large context window can lead to worse model performance

Reposted by mark thompson

PySport @pysport.org · 2h

We’re excited to launch the 𝐒𝐤𝐢𝐥𝐥𝐂𝐨𝐫𝐧𝐞𝐫 𝐗 𝐏𝐲𝐒𝐩𝐨𝐫𝐭 𝐀𝐧𝐚𝐥𝐲𝐭𝐢𝐜𝐬 𝐂𝐮𝐩 - a hackathon-style challenge that will bring together the community to collaborate, compete, and innovate using open-source SkillCorner data.

👉 Sign up pysport.org/analytics-cup

1 8 6

mark thompson @markposts.bsky.social · 16h

Also there should be book recycling boxes at supermarkets

1

mark thompson @markposts.bsky.social · 16h

Books should make more effort to give readers an idea of if they'll like them (bookshop recommendation tags from staff are often better indicators than blurbs, I think?)

Rachel Feder @rachelfeder.bsky.social · 1d

Tell me your most unhinged literary opinion, as a little treat

1 1 5

Reposted by mark thompson

mark thompson @markposts.bsky.social · 2d

The @statsbomb.com public dataset that I used for this is also *fascinating* for 'in-play vs exciting time' question, because you've got 'Big 5' teams, Messi Barca teams, and then NWSL, Indian Super League

The real mystery: what's up with Goa 2021/22?

www.getgoalsideanalytics.com/time-on-the-...

Time on the clock

How much time do teams spend in the attacking third? And do teams with more 'ball-in-play' time also have more 'final third' time in their games?

www.getgoalsideanalytics.com

1

Reposted by mark thompson

mark thompson @markposts.bsky.social · 2d

Scratched an itch by looking at whether there's a link between higher 'ball-in-play' time and higher 'worth watching'/'exciting' time

www.getgoalsideanalytics.com/time-on-the-...

Time on the clock

How much time do teams spend in the attacking third? And do teams with more 'ball-in-play' time also have more 'final third' time in their games?

www.getgoalsideanalytics.com

1 1 2

Reposted by mark thompson

Megan Feringa @meganferinga.bsky.social · 23h

Jess Park is very good at football. Apparently the only way to stop her is by kicking a football into her face, That’s after coming off the bench + scoring two goals. Given the freedom to invert, to carry the ball, to roam, Park is thriving. Man Utd are better for it. www.nytimes.com/athletic/671...

Jess Park’s sparkling cameo proves she is already integral to Manchester United

The brace against Everton from the bench was emblematic of Park’s transformation into one of the league’s most valuable attacking threats

www.nytimes.com

4 14

mark thompson @markposts.bsky.social · 1d

They don't code OGs

Kevin M. Kruse @kevinmkruse.bsky.social · 1d

In honor of spooky month, share a 4 word horror story that only someone in your profession would understand:

Archives banned making copies.

Natasha Simonova @philistella.bsky.social · 1d

In honor of spooky month, share a 4 word horror story that only someone in your profession would understand:

New library catalogue website.

1

mark thompson @markposts.bsky.social · 1d

Or, "this note-taking tool will sync basic descriptions of match action with notes you write (it adds lines that you can then edit if desired) so you can spend more of your attention on the match and your writing"

mark thompson @markposts.bsky.social · 1d

"offload 80% of the headspace you usually spend on the rote match action paragraphs 6-12, and spend more time on something unique and crafted for paragraphs 2-6, and we'll automate a bunch of out-linking to our other content so you/the subed doesn't have to" is where I'd start my product development

1

mark thompson @markposts.bsky.social · 1d

Stephen, she's twenty-fiv-- [remembers that Kim Little is still a starter] ok, fair

1 1

mark thompson @markposts.bsky.social · 1d

tbf I suspect that is more a case of being extra clear to readers about what they're reading, I think it's better to err on the side of clarity. I prefer that than potentially making it look like/making it unclear whether a person wrote it

1

mark thompson @markposts.bsky.social · 1d

This is doubly weird because I was always told that post-game write-ups were a (weirdly) persistent source of traffic for websites.

If you're a small outlet who doesn't have the resources to cover all games you want to, I see it, but as the league itself? With Apple money?

2 1 3

mark thompson @markposts.bsky.social · 1d

From a business and, more importantly, moral perspective I don't think you should ignore human media members to use gen AI (language or image gen) instead, but I would barely even call this a beta version of a product.

1 2

mark thompson @markposts.bsky.social · 1d

Another to add to the list of 'my assumptions about the direction of generative AI were wrong because established companies are putting out *bad products*'

Jeff Rueter @jeffrueter.bsky.social · 1d

As someone whose time freelancing for MLSsoccer was a vital step in my career progression, and someone who generally gleans insight from the observations of others:

This sucks so hard.

MLS is using generative AI to write match recaps. Writers used to get paid freelance rates to do that. Hssssss.

1 2 24

mark thompson @markposts.bsky.social · 2d

lol, cannot believe full-time was blown before taking the corner

1 1

mark thompson @markposts.bsky.social · 2d

It's absurd

mark thompson @markposts.bsky.social · 2d

This is badly worded really. Chance creation doesn't always mean that your chance creation will continue, and football is weird for the gap between 'good chance' and 'normal chance'.

If you get several one-v-ones and miss them all, as Villa have, that's Big xG that *could* just dry up

mark thompson @markposts.bsky.social · 2d

One of the quirks of football and stats is that failing to score means your chances are for nothing.

Villa put up 2.5 xG in the first half, missing four Big Chances, 0-0 at half-time, and have only just taken their first shot of the second half in the 70th minute (it was from ~22 yards out)

mark thompson @markposts.bsky.social · 2d

One of the quirks of football and stats is that failing to score means your chances are for nothing.

Villa put up 2.5 xG in the first half, missing four Big Chances, 0-0 at half-time, and have only just taken their first shot of the second half in the 70th minute (it was from ~22 yards out)

1 2

mark thompson @markposts.bsky.social · 2d

Really interesting article

www.nytimes.com/athletic/670...

Eight expectations about football from 25 years ago that have proven misguided

From pay-per-view coverage to the loss of technical midfielders to indoor stadiums – some expectations from 2000 haven't come true

www.nytimes.com

1

mark thompson @markposts.bsky.social · 2d

The @statsbomb.com public dataset that I used for this is also *fascinating* for 'in-play vs exciting time' question, because you've got 'Big 5' teams, Messi Barca teams, and then NWSL, Indian Super League

The real mystery: what's up with Goa 2021/22?

www.getgoalsideanalytics.com/time-on-the-...

Time on the clock

How much time do teams spend in the attacking third? And do teams with more 'ball-in-play' time also have more 'final third' time in their games?

www.getgoalsideanalytics.com

1

mark thompson @markposts.bsky.social · 2d

Scratched an itch by looking at whether there's a link between higher 'ball-in-play' time and higher 'worth watching'/'exciting' time

www.getgoalsideanalytics.com/time-on-the-...

Time on the clock

How much time do teams spend in the attacking third? And do teams with more 'ball-in-play' time also have more 'final third' time in their games?

www.getgoalsideanalytics.com

1 1 2

mark thompson @markposts.bsky.social · 2d

Should drop in a few hours

mark thompson @markposts.bsky.social · 3d

got a really fun ol' blog/newsletter cooking. subscribe at getgoalsideanalytics.com

2

Reposted by mark thompson

mark thompson @markposts.bsky.social · 3d

got a really fun ol' blog/newsletter cooking. subscribe at getgoalsideanalytics.com

2 1

Reposted by mark thompson

nilay patel @reckless.bsky.social · 2d

It was @sarahjeong.bsky.social and we’ve been pounding this drum for years - especially since all the promises of metadata to prove images are real have gone nowhere www.theverge.com/2024/8/22/24...

5 82 270