iiinova — where your digital self keeps evolving

Reposted by Matthew Johnson

Helen Kubiw ❤️ 🇨🇦 📚 @hkubiw.bsky.social · 05/10/2026

As readers, we already know the value of a library, but @jgcanada.bsky.social's new #MG #NF book THIS IS A LIBRARY BOOK (illus by Chanelle Nibbelink) from @kidscanpress.bsky.social can teach even a good reader more. Out tmrw, review today at canlitforlittlecanadians.blogspot.com/2026/10/this... ❤️👍🇨🇦📚

Book cover of "This Is a Library Book," subtitled "Check Out the Places and People Sharing Knowledge and Building Communities," written by Joyce Grant, illustrated by Chanelle Nibbelink, shows a circle of people linked by shared books.

054

Reposted by Matthew Johnson

Renee DiResta @noupside.bsky.social · 17h

This is a thing confirmed by research, to an extent. It’s also why ensuring content shows up in retrieval and in training data is the new battleground.

114131

Reposted by Matthew Johnson

Lukas Gunschera @lukasgunschera.com · 06/10/2026

Addiction has become the default framework for interpreting problematic social media use. It shapes how it is measured, discussed, regulated, and what support young people are offered. But does it match the experiences of young people who report feeling addicted to social media? 🧵(1/7)

35523

Reposted by Matthew Johnson

National Trust @nationaltrust.org.uk · 06/10/2026

You're never too old to get excited about seeing a really big tree.

291742434

Reposted by Matthew Johnson

Meredith Rose (from the grave) 👻 @mrose.ink · 05/10/2026

It’s a frustrating case, but I have yet to see any theory of “not a fair use” that doesn’t completely obliterate most extant fair uses in its wake.

1103

Matthew Johnson @matthewfjohnson.bsky.social · 05/10/2026

Even if those are true (& I think some of them are) they don't retroactively create environmental costs. Saying "AI MUST have high environmental costs because AI is bad" or "environmental harms of AI are worse than things that use way more water and pwoer because AI is bad" is motivated reasoning.

100

Matthew Johnson @matthewfjohnson.bsky.social · 05/10/2026

Comparison to zero is the same as the absolute measure ("Is that a lot?"). In my opinion that is highly relevant on the local scale but not super relevant on the macro scale, where "Compared to what?" becomes more important.

120

Matthew Johnson @matthewfjohnson.bsky.social · 04/10/2026

Always fun to see someone reach the "retreat and claim victory" stage

000

Matthew Johnson @matthewfjohnson.bsky.social · 04/10/2026

I agree completely! That's what makes me wonder why people are so fixated on AI in this context.

000

Matthew Johnson @matthewfjohnson.bsky.social · 04/10/2026

It just seems to me that if we find we're applying different standards to similar things, we should reflect on why we're doing that, instead of doubling down on our current beliefs.

100

Matthew Johnson @matthewfjohnson.bsky.social · 04/10/2026

Yes, that's precisely the one I pointed out was comparing apples to oranges.

200

Matthew Johnson @matthewfjohnson.bsky.social · 04/10/2026

Again, though, relative to what? It's reasonable to be concerned about a dripping faucet, but not if your house is in fire. I haven't seen any evidence that AI's environmental impact is significantly greater than or even equal to other industries (& uses of data centres) we mostly give a pass to.

100

Matthew Johnson @matthewfjohnson.bsky.social · 04/10/2026

I think there's a middle ground between "expecting everyone to be data driven" and "basing the debate entirely on vibes," which is where we are now.

100

Matthew Johnson @matthewfjohnson.bsky.social · 04/10/2026

No, but it's a reason not to compare estimates of AI's environmental impact that do take a cradle-to-grave approach to estimates of other industries' impact that don't take that approach.

130

Matthew Johnson @matthewfjohnson.bsky.social · 04/10/2026

Well, steel is one of the big missing pieces from our environmental analysis of agriculture, whether we're talking about water, electricity or carbon.

130

Matthew Johnson @matthewfjohnson.bsky.social · 04/10/2026

Yes, exactly - we treat them differently for reasons that are all about our perception and not the data. Not a great basis for policy or public discourse!

100

Matthew Johnson @matthewfjohnson.bsky.social · 04/10/2026

If you look at the section on defining water footprint, you'll see they argue for taking a broader approach than is currently taken, which shows that current water footprint measurements don't take a cradle-to-grave approach.

120

Matthew Johnson @matthewfjohnson.bsky.social · 04/10/2026

The corrolary of "everything bad people do is bad" is "everything good people (eg me) do is good," which is a very dangerous attitude.

100

Matthew Johnson @matthewfjohnson.bsky.social · 04/10/2026

And I think the difference in proportionality is important because it suggests people are engaging in motivated reasoning - concluding that AI is bad for the environment _because_ of the other things you named, not due to actual data - which IMO is always a bad basis for policy.

000

Matthew Johnson @matthewfjohnson.bsky.social · 04/10/2026

Meat is not necessary for people's lives. Under the best conditions it's a way of turning indigestible cellulose into digestible food, but we feed livestock soy and corn - food humans CAN eat - in a process this much more wasteful than AI could possibly be.

200

Matthew Johnson @matthewfjohnson.bsky.social · 04/10/2026

Here's an analysis of water footprinting that shows the cradle-to-grave approach is not generally taken in analysing industry impact: assets.wwf.org.uk/downloads/wa...

130

Matthew Johnson @matthewfjohnson.bsky.social · 04/10/2026

My point is that the cradle-to-grave analyses of AI costs _are_ including everything while analyses of other industries aren't. It's apples to oranges.

140

Matthew Johnson @matthewfjohnson.bsky.social · 04/10/2026

Similarly there's some discussion of how recently built data centres frequently use standalone power plants, which are usually much less efficient and more polluting than the local grid, but little acknowledgement that this is often done to assuage local concerns about power costs going up.

110

Matthew Johnson @matthewfjohnson.bsky.social · 04/10/2026

That said, I think there are absolutely legitimate environmental concerns about data centres at the local level. WHERE one is built - how the grid's electricity is generated, how much water is available - are key questions. They don't seem to be a big part of the conversation though.

110

Matthew Johnson @matthewfjohnson.bsky.social · 04/10/2026

Which is not to say that AI, and the AI industry in particular, aren't bad - they are, in many ways! Nor is it wrong for anyone to be concerned about them. But when, as a community, our concern is disproportionate and also moralized in ways similar issues aren't, I think it's worth asking why.

210

Matthew Johnson @matthewfjohnson.bsky.social · 04/10/2026

In this case, I think it's mostly a question of salience: people don't realize that most North American grain is raised to feed to animals, adding a completely unnecessary and insanity inefficient step; and they hear constantly the AI is bad, so they're primed to accept anything that supports that.

1111

Matthew Johnson @matthewfjohnson.bsky.social · 04/10/2026

IMO a big part of critical thinking and intellectual humility is asking ourselves why we think something, and testing that to see if what we think matches reality. Proportion is a big part of that.

110

Matthew Johnson @matthewfjohnson.bsky.social · 04/10/2026

Again, though, it's "compared to what"; we rarely consider the impact of other industries in a cradle-to-grave way - cost estimates for agriculture, for example, don't generally include the cost of making the steel for farm equipment. To do so only for AI is special pleading.

4202

Reposted by Matthew Johnson

AgonyFlips @agonyflips.bsky.social · 20/09/2026

When you're ticked off and a friend sends you something to make you feel better. 🐑🔊

493157873846

Matthew Johnson @matthewfjohnson.bsky.social · 03/10/2026

It's like how we worry more about plane crashes than car crashes, or how we weighed the risks of nuclear power vs the costs of continuing to rely on fossil fuels.

030

Matthew Johnson @matthewfjohnson.bsky.social · 03/10/2026

It isn't, it's a question of proportion ("Compared to what?") If our level of concern about something is out of proportion to its relative importance, that doesn't mean we shouldn't worry about it at all, but it does mean we should think about why we are so worried about this in particular.

230

Matthew Johnson @matthewfjohnson.bsky.social · 03/10/2026

Right, but are we twice as concerned about water/power use for AI as we are about streaming Video, or are we 1000x times more?

130

Matthew Johnson @matthewfjohnson.bsky.social · 03/10/2026

... and also, how big our concern about it is relative to those other things. If we're focused more heavily on something that is a much smaller issues, it suggests to me that we're actually worried about something else and water/power is just a proxy.

140

Matthew Johnson @matthewfjohnson.bsky.social · 03/10/2026

I agree it doesn't mean it isn't an issue - and as I said, some uses of AI are way more energy-intensive than others. The question is how big of an issue it is relative to either a) other things that use a lot of water/power and b) other things we do with data centres...

130

Matthew Johnson @matthewfjohnson.bsky.social · 03/10/2026

Independent analysts have found similar figures: towardsdatascience.com/chatgpts-ene...

020

Matthew Johnson @matthewfjohnson.bsky.social · 03/10/2026

In each case, though, we don't tend to think about how much water or power we're using when we do any of those things (though I hope people think about it before eating meat, especially beef.) And some things are probably not worth doing with AI, like making video (likely why Sora was shut down).

020

Matthew Johnson @matthewfjohnson.bsky.social · 03/10/2026

Similarly, other things we do with computers - like streaming video or online meetings - use significantly more energy than an LLM. But here too "compared to what" is important - making images with AI use more energy, and making video uses A LOT more. www.forbes.com/sites/johnko...

241

Matthew Johnson @matthewfjohnson.bsky.social · 03/10/2026

Of course, it's fair to say that we need agriculture, and we don't need AI. But the sector of agriculture that takes the most water is raising livestock - mostly from raising soy, corn, alfalfa, etc. used for animal feed. If we want to use water more efficiently, that would be a better start.

261

Matthew Johnson @matthewfjohnson.bsky.social · 03/10/2026

Sure, this article has a good overview and links to its sources: www.outsideonline.com/outdoor-adve... "All U.S. data centers combined—not just AI, all of them—account for roughly 0.3 percent of total national water withdrawals. Agriculture consumes approximately 80% of Colorado River water."

362

Matthew Johnson @matthewfjohnson.bsky.social · 03/10/2026

There are two basic questions to ask about statistics, "Is that a lot?" and compared to what?' On AI and water use, the answers to the first is "Yes," but the answer to the second is "Not compared to other products and industries whose water and power use we're not usually worried about."

170

Reposted by Matthew Johnson

Isaac Bogoch @isaacbogoch.bsky.social · 02/10/2026

Respiratory virus activity remains low in Canada, but is starting to rise. Vaccine programs are now rolling out. 🇨🇦

3144

Matthew Johnson @matthewfjohnson.bsky.social · 30/09/2026

It's a little less weird than it seems. That text is adapted pretty closely from the "End Poem" that appears when you "win" Minecraft: en.wikisource.org/wiki/End_Poem

en.wikisource.org

End Poem - Wikisource, the free online library

100

Matthew Johnson @matthewfjohnson.bsky.social · 30/09/2026

It's a little less weird than it seems. That text is adapted pretty closely from the "End Poem" that appears when you "win" Minecraft: en.wikisource.org/wiki/End_Poem

010

Matthew Johnson @matthewfjohnson.bsky.social · 30/09/2026

Fiction: Everyone speaks the Common Tongue Reality: A medieval person from England was jailed after being shipwrecked in another part of England, because the people there couldn't understand him and assume he must be a foreign spy

011

Matthew Johnson @matthewfjohnson.bsky.social · 25/09/2026

Or the Alan Moore version where Mary Poppins kills Harry Potter.

010

Matthew Johnson @matthewfjohnson.bsky.social · 25/09/2026

The problem with watching everything with captions on is I have trouble remembering which shows actually weren't in English

010

Matthew Johnson @matthewfjohnson.bsky.social · 25/09/2026

Lupin of course. Somebody else already mentioned Dark.

110

Matthew Johnson @matthewfjohnson.bsky.social · 24/09/2026

Alas texting was also pretty strongly associated with anxiety and depressive symptoms.

110

Matthew Johnson @matthewfjohnson.bsky.social · 24/09/2026

Here it is: Mougharbel, F., Chaput, J. P., Sampasa-Kanyinga, H., Colman, I., Leatherdale, S. T., Patte, K. A., & Goldfield, G. S. (2023). Longitudinal associations between different types of screen use and depression and anxiety symptoms in adolescents. Frontiers in Public Health, 11, 1101594.

110