34 Comments
User's avatar
Stuart Henderson's avatar

I am reminded of Marshall McLuhan.

Expand full comment
Rohit Krishnan's avatar

Yes it's definitely echoed, though in a larger scale as it's not just the medium changing the message but the topology of the medium connecting us that changes the messages which gets boosted.

Expand full comment
Ben Reid's avatar

Hugely insightful, extremely well structured and measured analysis. Thank you Rohit🙏

Expand full comment
Rohit Krishnan's avatar

Thanks for the kind words!

Expand full comment
Étienne Fortier-Dubois's avatar

This is great! I particularly enjoyed the point about microcultures and the macroculture. It's easy to make a point that the general trend is towards either of these, while missing the subtle dynamics that are at play. Maybe both microcultures and the macroculture are becoming more powerful just because, thanks to the density you describe, there's *more* culture overall.

Expand full comment
Rohit Krishnan's avatar

Yes! It's both, and the nuances are what's interesting I think. There's more microcultures but discovery is hard and you end up feeling like it's a set of fleeting microcultures than anything with a sense of permanence. A feeling of standing amidst flux.

Expand full comment
Alex's avatar

Thanks for your thoughtful post. I enjoyed a lot of it, and have it bookmarked as a reference to share with others who way to learn more about networks.

A few updates could make this more impactful, however.

First, for complex contagion to spread, you want locally dense network and not globally dense networks, otherwise noise will cause contagions to die out before they reach a critical threshold – you kinda get at this a bit when you talk about there being "less robustness and stability" as density increases:

* https://www.journals.uchicago.edu/doi/abs/10.1086/521848

* https://www.sciencedirect.com/science/article/abs/pii/S0378873312000111

* https://www.researchgate.net/publication/288180127_Complex_contagions_and_the_diffusion_of_popular_Twitter_hashtags_in_Nigeria

* https://ieeexplore.ieee.org/document/9381313

Section IV misrepresents the effects of density because it doesn't differentiate between where the density is. See also

* https://www.youtube.com/watch?v=O8LAZylisRk

* https://www.youtube.com/watch?v=bTXUJQhEqL0

For example, what you stated in Section IV is true for simple contagion but not for complex contagion, and most of what you're discussing is complex contagion. It's a bit confounding and misleading to bounce between talking about how ideas become contagions and then how power grids go down – the former is a complex contagion but the latter is a simple contagion. They're not the same. A power grid going down doesn't need repeated exposures to something to go down; however, adopting a new idea or behavior does need repeated exposures – and often from a variety of people – before it gains enough legitimacy, credibility, coordination, and emotionality to spread virally in a network (not to mention other challenges of contested vs non-contested ideas/behaviors): https://en.wikipedia.org/wiki/Complex_contagion.

Also, most of these examples of complex contagions are affected by multiplex dynamic networks (https://youtu.be/bTXUJQhEqL0?si=JBCek-IyUfU_jTT0&t=410). In a few places above, you discuss culture and networks people find themselves in as one thing vs a plurality of positions one occupies simultaneously. In some cases, you skirt this by talking about macroculture and microculture, but this this confuses points you made above when you speak about the one big great ocean of information culture we're all in. For readers who can't sift things out, it may mislead or confuse them.

Lastly, it's a bit circular to work around this by saying that the contagions we observe were "only those that could spread," and this point goes against your larger point that I love which is that the structure of networks has a huge impact on what will become a contagion and what will die out. See here for an article critiquing the idea that "only the narratives which can go viral do go viral": https://www.science.org/doi/10.1126/science.1121066.

All that being said, I really did enjoy this and hope you consider some of the revisions I suggested to make this clearer, more precise, and more impactful.

Expand full comment
Rohit Krishnan's avatar

Thank you for the wonderful thoughtful comment. I’ll read the links, and comment!

Expand full comment
Dahlia Daos's avatar

Is there a way to interact with the network with intention? Can you "drive" how the network reacts and create pressure towards certain kinds of topology? If you were to apply enough pressure in enough points, could you ultimately change how information spreads and prevent dense clusters from forming?

Also, do you know of Geoffrey West's power laws about energy use? He calls it the fourth law of thermodynamics. I don't understand anything on the information side of entropy, but is there a similar kind of power law at work here? Is information density a measure of entropy?

Expand full comment
Rohit Krishnan's avatar

You should be able to do this through controlling the information flow and centrality of nodes. Arguably this is what we already do unintentionally. And yes it has echoes of Geoff West's theory. Thinking of information density as a measure of entropy is really intriguing.

Expand full comment
Dahlia Daos's avatar

Also, can you recommend a good primer on networks?

Expand full comment
Dahlia Daos's avatar

What would intentional control of information flow and centrality of nodes require/imply?

Expand full comment
Rohit Krishnan's avatar

I learnt from Duncan watts and geoff west Santa Fe papers primarily iirc. And that's the main media and culture narrative power here, as one example.

Expand full comment
Paul Humphries's avatar

Of course there's also the complication that we're not just nodes but apes, and that each ape is unique, and we don't yet understand how DNA affects our behavior, nor do we understand how the brain works.

Expand full comment
Derek Beyer's avatar

I'm really curious about how this interacts with geography. So much of our networking once took place in meatspace and potential connections were largely defined by proximity. Mail, newspapers, radio, and television certainly affected that, but I'd imagine that it mainly supported more scale rather than divorcing networks from geography entirely.

Expand full comment
Rohit Krishnan's avatar

Geography helps create active separation in real life, but in the digital world there are no separations

Expand full comment
Huvinesh Rajendran's avatar

What a great read ! Keep up the good work.

Expand full comment
ProtopiacOne's avatar

Excellent problem statement! Any hope of design-building some solutions, or better to just wait to see what emerges?

Expand full comment
Rohit Krishnan's avatar

If you have ideas ...

Expand full comment
ProtopiacOne's avatar

I had some ideas... but then AI came around and changed the calculus on ideas...

Expand full comment
Angela Natividad's avatar

This is so orderly and satisfying

Expand full comment
Rohit Krishnan's avatar

Thank you! That's the highest praise .

Expand full comment
Scott Werner's avatar

This post couldn’t be more timely! I’ve recently been feeling and discussing similar ideas with people around software development.

We’ve noticed there isn’t really a “counter culture” anymore, and I really like the dark forest theory that if you connect to the wider internet too soon before the ideas have had a chance to mature, it dies out.

One illustrative example I think is the collective outrage of the typescript community when DHH decided to move off of it in a library he was building (that most didn’t even use). He’s already in a position he can weather the attacks, but some new community probably not so much.

I guess if people start breaking away into these cultural islands, if they’re able to sustain long enough to let their ideas mature and break out, we’ll see much more fully formed, groundbreaking things coming from them and not really seeing the intermediate products unless you explicitly seek them out?

Expand full comment
Chris Fedrizzi's avatar

Shit brother. I am too stupid for this essay yet. Where should I get started to learn about networks? Much appreciate the references, if you have suggestions.

Expand full comment
Rohit Krishnan's avatar

Oh man I'm sorry, that's a failure of writing on my part. Maybe this is good? https://meltingasphalt.com/interactive/going-critical/

Expand full comment
Chris Fedrizzi's avatar

Thanks brother. I don't think it's your writing, I just need to do the homework before reading it again. I'll learn more about networks and come back here soon. Thanks!

Expand full comment
Chris Fedrizzi's avatar

Thank you Antony! I am diving deep into the pdf. Really good stuff.

Expand full comment
Antony Turner's avatar

Glad you're enjoying it. There is a free video course too, which is really good.

https://www.edx.org/learn/computer-programming/cornell-university-networks-crowds-and-markets

Expand full comment
Priscilla Estes's avatar

Very thoughtful. Thank you. My angst level decreased once you defined the problem.

Expand full comment
Rohit Krishnan's avatar

Thank you!

Expand full comment
Hollis Robbins (@Anecdotal)'s avatar

Fantastic essay -- thinking of course of Ferdinand Tönnies's great work on Gemeinschaft and Gesellschaft (1887) which has been so influential and prescient. We make and remake our networks. https://en.wikipedia.org/wiki/Gemeinschaft_and_Gesellschaft

Expand full comment
Rohit Krishnan's avatar

Thank you! I'm not familiar with it, will read!

Expand full comment
ProtopiacOne's avatar

Excellent problem statement! Any hope of design-building some solutions, or better to just wait to see what emerges?

Expand full comment