iiinova — where your digital self keeps evolving

The People do not yearn for automation, but idk there's some sort of thing that's happening

man, I got Claude to take a screenshot of my updated personal analytics library app to put on my personal site, and then was like 'can you try blurring some of that so papers and authors aren't identifiable' and it just did it, far more precisely than I thought it would

Simon Willison @simonwillison.net · 28/09/2026

Reposted by mark thompson

I really like how @reckless.bsky.social described this: "The people do not yearn for automation" www.theverge.com/podcast/9170...

theverge.com

BEWARE SOFTWARE BRAIN

Software brain is changing the world, but most people still aren’t buying.

5728

Gonna contrive myself into tapping my 'Twitter Circles were good and bsky should do them' sign, because half these quote posts and comments would be Circle'd and stop the spread of context collapse

In the meantime though, chart looks good bsky.app/profile/nath...

Reposted by mark thompson

All data is wrong, some of it is useful

'All data is wrong, some of it is useful'; actual established methods of thinking about data quality; and pandas www.getgoalsideanalytics.com/all-data-is-...

getgoalsideanalytics.com

131

yeah, I'm really intrigued about whether the various (soccer) football attempts have struggled to take off because of the *sport* or because Hanson is the Tom Brady of that particular kind of presenting

You could maybe throw in the spread of NFL to other TV windows (and maybe to international games too) while preserving high school and college football

000

Is there a book about NFL RedZone? Feel like there'd be a really great book in it about TV production, technology, and sports culture

210

Arsenal had a pretty good hold on the match so conceding like that's gotta sting Meanwhile, a lot of Chelsea's threat has been 'Alyssa Thompson goes vroom', so it's fitting she got the goal

050

4-0, 4.4 xG, 34 shots, 70 touches in the box.

Their October run of games is going to be really tough and really interesting: London City (H) Chelsea (A) Arsenal (H) Arsenal play Man City in that timespan, and City also play United. Lot of big points could swing a lot of ways

120

Spurs had a goalmouth scramble cleared off the line at 0-0 and have not let the foot off the gas since. 3-0 up against Villa (they'll go 2nd in the WSL), a statline of 31 shots to 6, 4.05xG to 0.44

160

Never let them know your next move 🐼

020

All data is wrong, some of it is useful

'All data is wrong, some of it is useful'; actual established methods of thinking about data quality; and pandas www.getgoalsideanalytics.com/all-data-is-...

getgoalsideanalytics.com

131

Reposted by mark thompson

Doug Gordon @brooklynspoke.bsky.social · 26/09/2026

Some rules I've tried to teach my kids about social media and media in general. Before reacting, ask: 1. Why do I know this? 2. Is it important? 3. What is the post/story *not* showing me? 4. Does anyone actually need my opinion on this? I don't always live up to these myself, but they're helpful!

19013

Immediately they concede a free-kick close and central, which Alex Greenwood scores. Good posting.

000

Charlton have been pretty good at sending individual pressers out of their organised defensive block. Always kinda worry about how long that energy and concentration will be able to last, but gotta do what you gotta do

They've got drone shots now, they should put them to good use with this

American Soccer Insights Summit @americansoccerinsights.com · 25/09/2026

F1 comms needs to take a note from Tour de France commentary. What are the regional speciality dishes? What is the history of the city?

320

Reposted by mark thompson

Interested in presenting research at the ASI Summit? We are opening a call for proposals, with data provided by SkillCorner. For more details, visit: americansoccerinsights.com/call-for-pro...

Call for proposals, deadline: October 23. In partnership with SkillCorner.

055

mark thompson @markposts.bsky.social · 23/09/2026

Wanting to understand how the model works more than how the output usage works, which is what I've seen more of. I've seen 'it makes decisions instead of text', but I wanna know how

mark thompson @markposts.bsky.social · 23/09/2026

What's the best explanation y'all have seen about how Jev actually works?

200

Reposted by mark thompson

cata @catabush.com · 23/09/2026

Morning newsletter for your perusal. No closer look as I am lacking inspiration this Wednesday morning. Plenty of interesting MLS updates, though:

post-match-report.leaflet.pub

September 22: Updates

Lots of MLS news

093

As a(nother) general rule, I think it benefits media folk to be educated on the biz and the modern biz is partly a tech biz (partly/largely because the distribution biz, which was always part of the media biz, is now so heavily techy)

To be clear: I am slightly worried about bluesky's tendency of interpreting 'this is a factor someone may have considered' as 'this is my firm and deeply-held opinion'.

100

But, in media, a factor in 'is this a good product' *is*, actually, 'will our audience care if a person is earning money for this' Again, not talking Goal specifically, but it's very possible that 'better content' might be a worse product if it erodes brand trust (*also* true of 'SEO-slop')

100

A lot of these arguments could also be made about 'SEO-slop' pieces (what time *is* the Super Bowl). Who is it for? (Apparently people did/do click those pieces) Why do that instead of X? (the clicks brought in ad revenue to do other things)

100

But what counts as 'good product' will vary by outlet. For outlets closer to specific clubs, e.g. local papers, a stronger understanding of the specific club & their history is the minimum acceptable standard; or e.g. for a national outlet, that nation's players (Scottish outlet covering Serie A)

100

On development, I think a lot of people outside product biz aren't familiar with what development is like. If the up-front cost is high, you might decide to make use of it in stages to either test or otherwise justify cost, rather than eating cost by testing internally.

100

That all means that if the product is decent, you can be more responsive to the moment and drag in low-cost* ad revenue that you basically bank to - **the key thing** - invest in other projects. *(On an ongoing basis; development of the tech is an up-front cost, of course)

100

Additionally, I imagine, not a totally direct, 1:1 comparison with 'why not hire a human'. I think(?) that the editorial coverage decisions for regular player ratings have to be made pre-match; and I also assume editorial decisions for this type of article could be made ~anytime before full-time.

100

At the same time, everyone kinda knows that traditional player ratings are rushed and narrative-driven anyway. It's not necessarily clear that it's the journalist's stamp on player ratings that leads to clicks so much as it being 'category: post-game chat'. Why this matters is further down

100

My general starting point is 'how does techifying stuff help people do better work, or at least keep the lights on' Competition is fierce. All your favourite score apps have 'AI' player ratings *anyway*, and many other apps/outlets have stats or highlights or both. That's the landscape for eyeballs

100

The Mill @manchestermill.bsky.social · 21/09/2026

This is sort of sideways related to Goal specifically, but as someone whose career has spanned media and tech, want to at least broaden out some thoughts This isn't a thread with any particular opinion about the feature, it's a very 'take what you want from this' thread

100

Reposted by mark thompson

We have two weeks to reach a huge new milestone: our first 5,000 paying members. We have around 450 to go. We’re currently running a discount of just £6.95 a month for your first two months, or, hugely: Just £69 for a whole year. manchestermill.co.uk/more-than-ne...

manchestermill.co.uk

More than news: Help us get to 5,000 Millers

Join our new campaign!

0811

mark thompson @markposts.bsky.social · 20/09/2026

Thing with xG is that it generally smooths out the luck/variance-based noise inherent in goal-scoring, but xG itself still has room for big trowels of luck

0233

mark thompson @markposts.bsky.social · 20/09/2026

Now that the current season is up and running, I'll probably add a mode where you can roll for 2026/27 seasons. The (interesting but awkward) bit is that you need a time period to estimate on, but realistically an attacker's game-time in a season is linked to their goal-scoring