As VC’s become quasi-media orgs and AI leverage augments our ability to produce, it’s increasingly hard to keep up. Signal to noise is degrading, making taste and careful curation of feeds a must. Not just to avoid falling behind, but to filter out the slop and chase uncovered intellectual alpha neglected by the crowd.
My personal feed is a combination of (a) deliberate content (~200 substacks/blogs, ~10 podcasts), (b) timeline serendipity curated based on a few hundred interesting nodes and (c) networked content as a product of (a) & (b) which is hopefully out-of-distribution but hallmarked with credibility. Like everyone, it’s an overwhelming volume and in response I’ve noticed some emerging patterns in how I engage:
- Full read // The Library. Trusted authors whose content often contains truly unique insights and / or thought patterns I need to reason through. Unsurprisingly these publications tend to be paid, have full-time writing authors or come from hyper-legible ecosystem nodes. Though the latter is often a weak filter.
- Convert to audio // The Treadmill. I know this is a more lossy format, but I’m busy and the content is usually overly verbose. I’ve gained value from their writing before, but I’m not convinced the author has a unique insight on this topic. Anyway, it feels more productive than Netflix, especially when paired with exercise.
- LLM summary // Canteen Food. In this bucket goes the bad writers, experts with a finger on the pulse but transfixed to some boring meta agenda (e.g. Roy Lilley) and ‘forced content’, those publications that are ‘in-network’ with the crowd so influence peoples priors but kinda suck (e.g. MSM). Finish your plate son.
- Block. Two strike rule. If somebody serves up a bucket of LLM slop you are outta here.
The opportunity cost of time makes curation king, but it can also narrow your aperture. The more refined your filters, the more you miss weak signals of something novel. Where I’m already well informed, triage friction is lower. Comparatively the inertia in deciding to engage with what may be sources of out-of-distribution ideas increases. It’s a broader macro question: how to deal with trust in a low trust world?
I deeply respect those like Josh Wolfe who actively seek out weird information as part of their daily routine. If Hollywood is our bellwether, the future is infinite Marvel sequels and remakes. Safe bets, low variance, and an easy default of being a passive indexer of the timeline. But can you risk outsourcing your taste when little else may matter? The challenge (and maybe the edge?) is cultivating routines that keep friction alive