woensdag 13 augustus 2025

Sora experiment: Make a video of a Mountainbiker racing downhill on a slippery windy track next to a very steep and high cliff in the rain

Make a video of a Mountainbiker racing downhill on a slippery windy track next to a very steep and high cliff in the rain

Fat tails vs. AI agents 6-9 years from buzz to reality

how much time was there between the buzz about fat tails due to the internet and actual fat tails becoming visible like 1000's of niche youtube channels instead of a few tens of tv channels and media outlets before

----

If we look at the historical sequence, the "buzz" about fat tails and the internet—often called the long tail effect—started in the early 2000s, especially after Chris Anderson's 2004 Wired article The Long Tail and his 2006 book of the same name. At that point, the idea was that digital distribution would allow huge numbers of small, niche products and creators to coexist alongside mass hits, rather than just the few blockbusters dictated by scarce shelf space or limited TV channels.

Timeline breakdown:

  • Late 1990s–early 2000s – The internet's reach and early platforms (eBay, Amazon, blogs) were already making niche markets possible. People started noticing that obscure books, indie music, and niche communities could thrive online, but the phenomenon wasn't fully visible at scale.

  • 2004–2006 – The long-tail / fat-tail discussion became mainstream in tech and media circles. The argument was mostly theoretical, backed by some early data from Amazon, Netflix DVD rentals, and iTunes.

  • Mid-to-late 2000s – YouTube (2005 launch) began lowering barriers for video distribution, but the visible fat tail was still limited because broadband, monetization tools, and recommendation algorithms weren't mature.

  • Early 2010s – Broadband and smartphone penetration plus social media distribution made the tail truly visible: thousands of niche YouTube channels gained dedicated followings, podcasts exploded, and self-publishing surged. This is when the "fat tail" was obvious to everyday users, not just analysts.

  • Late 2010s onward – Algorithm-driven discovery and monetization solidified the fat tail as a permanent media structure. Many niches now rival or exceed mid-tier TV audiences from the pre-internet era.

Gap between theory and full visibility:
From the buzz phase (circa 2004–2006) to the obvious reality phase (circa 2012–2014), you're looking at roughly 8–10 years before the "thousands of niche channels" phenomenon became undeniable at scale. Early signs were there before that, but it took the maturation of platforms, recommendation engines, and monetization ecosystems for the fat tail to fully materialize.

If you want, I can also map this to specific tech and social milestones that triggered each stage so we can see exactly why it took that long. Would you like me to do that?

----

I suspect there will be a similar time gap between the current buzz about AI agents and them becoming a reality. your thoughts?

----

Your analogy makes sense, and the fat tail case offers a good framework for thinking about AI agents.

Here's how I'd break it down.

1. The "Buzz" Phase

Right now, AI agents are in roughly the 2004–2006 long tail stage.
We've got:

  • Conceptual hype – People are talking about autonomous agents that can plan, reason, and execute complex multi-step tasks with minimal supervision.

  • Early prototypes – AutoGPT, BabyAGI, and crew.ai show promise but are mostly brittle demos.

  • Excitement from insiders – The tech press, AI labs, and early adopters are enthusiastic, but the general public isn't interacting with true agents daily—yet.

The analogy: this is like when Chris Anderson was publishing The Long Tail, and everyone in media was saying "This will change everything," even though most people still watched the same few network shows.

2. Why There's Likely to Be a Delay

A genuine AI agent economy will require multiple layers to mature:

  • Core capability – Current LLMs can plan in short bursts but often hallucinate, lose track over long horizons, and lack robust real-world grounding.

  • Integration – Agents will need reliable hooks into APIs, data sources, robotic systems, and secure execution environments.

  • Economics – Platforms for monetizing, distributing, and controlling agents will have to emerge (like YouTube Partner Program for creators).

  • Trust & regulation – People will need to trust agents to act without constant oversight; governments and industry will likely set rules before mass adoption.

These bottlenecks mirror the delays we saw in the long-tail shift:

  • Tech existed (internet),

  • but infrastructure, interfaces, and trust took another decade.

3. Possible Timeline

If the analogy holds:

  • Buzz phase: 2023–2025

  • Early visible wins: 2026–2028 — Certain domains (e.g., customer service triage, internal research assistants) will have working autonomous agents.

  • Obvious mainstream reality: 2029–2032 — Consumers and businesses will interact daily with thousands of specialized agents (analogous to niche YouTube channels in 2012–2014).

That's ~6–9 years from buzz to "everyday reality," assuming tech progress stays on current curves.

4. Caveats

  • Acceleration possible – AI development cycles are faster than early internet cycles; model improvements, open-source contributions, and capital inflows could compress the timeline.

  • Or… stall – Bottlenecks in safety, reliability, or energy costs could stretch the gap longer.

  • The key difference from the long tail:
    The internet didn't face as many existential safety debates or hardware scaling constraints. AI agents do.

If you want, I can chart a side-by-side comparison of the long-tail timeline vs. a projected AI agents timeline so you can see exactly where we might be headed. That would make the analogy even sharper.


zaterdag 2 augustus 2025

Watchtower LiDAR. Dominant design?


The LiDAR in the "watchtower"-style dome behind the windshield roof line on the latest aito and xiaomi cars fascinate me. I wonder if they will become a "dominant design" (schumpeter). If so it would be one of the early cases in modern history where chinese manufacturers set the standard… my experience is that until now western brands set the standard and Chinese manufacturers copied it. Would be a bold statement by China

——

You've put your finger on a really significant shift in automotive history — and your intuition is spot on. The watchtower‑style LiDAR dome that Huawei (via AITO) and Xiaomi are now standardizing could indeed evolve into a "dominant design" in the Schumpeterian sense. Let's unpack this.





🏛️ What a "Dominant Design" Means (Schumpeter)



In innovation theory, a dominant design emerges when a particular configuration of a technology becomes the industry standard:


  • It balances performance, cost, and reliability well enough that competitors converge on it.
  • Examples:
    • QWERTY keyboard → standardized typing layout.
    • iPhone's slab touchscreen form factor → redefined the smartphone.
    • Tesla's skateboard EV platform → adopted across the EV industry.



So your question is: could the roof‑mounted LiDAR dome on Chinese EVs become the equivalent for autonomous mobility?





🚘 Why the LiDAR Dome Matters



  1. Technical Performance
    • Elevated "watchtower" placement provides a 360° panoramic field of view with fewer blind spots.
    • Reduces occlusion from bodywork, critical in dense urban driving.
    • Works especially well for long‑range LiDARs (200–250 m), essential for high‑speed safety.

  3. Design Convergence in China
    • Huawei (AITO M9, M8) → roofline LiDAR as signature.
    • Xiaomi (SU7 Pro/Max, YU7) → Hesai AT128 LiDAR in the watchtower position.
    • NIO, XPeng, Li Auto → many also adopting elevated roof LiDAR placement.
    • This clustering creates network effects: suppliers scale up, costs fall, and consumer recognition rises.

  5. Contrast With Western Brands
    • Tesla: vision‑only, refuses LiDAR.
    • Mercedes, BMW, Audi: experimenting, but LiDAR often hidden behind grilles or bumpers (lower vantage points).
    • Waymo & Cruise: use roof LiDARs on robotaxis, but not yet in consumer vehicles.
    • China is the first to normalize visible LiDAR domes in mass‑produced consumer EVs.






📊 Could This Become a Dominant Design?



  • Yes, if Level 3–4 autonomy becomes a regulatory and consumer expectation. A high‑mounted LiDAR provides safety redundancy that regulators may favor.
  • If Chinese automakers continue their global expansion, the roof LiDAR dome could become synonymous with "serious autonomy" in the same way that Tesla's flush handles signaled EV modernity.
  • Cost trends help: solid‑state LiDAR units have dropped from >$7,000 a decade ago to <$500 in 2025, making widespread adoption feasible.






🌏 Historical Significance



You're right: historically, Western brands set the template (Ford with assembly line, VW with affordable cars, Tesla with EV layout). Chinese manufacturers often iterated or copied.


But now:


  • China is setting a potential global template for autonomous‑ready vehicles.
  • If the watchtower dome is the dominant design, it will mark one of the first modern cases where China exports not just cars, but a design standard.
  • It would be a bold statement: instead of following, Chinese automakers lead the form factor of autonomy.






🧭 My Take



I think we're watching a paradigm shift. Unless Tesla's camera‑only approach dramatically outperforms, the roof‑mounted LiDAR dome could easily become the QWERTY keyboard of autonomous cars. And if so, this will go down as one of the first major instances of China dictating global automotive design norms.




⚡Would you like me to create a visual timeline of dominant designs in auto history (Ford → Tesla → Chinese LiDAR domes) so you can see how this fits into the bigger Schumpeterian cycle?


Hypothetical Fiscal Dependency Ratio (FDR), 2020–2100




Here's a hypothetical Fiscal Dependency Ratio (FDR) projection for China, Europe, Japan, the USA, and Africa (2020–2100), assuming an elderly dependent costs ~2.5× as much as a youth dependent.



Key Takeaways



  • Japan: Already very high; by 2100, one worker may effectively bear the fiscal cost equivalent of 2+ dependents.
  • China: Huge rise, overtaking Europe and approaching Japan's level by 2100.
  • Europe & USA: Significant increase, but less extreme than China/Japan thanks to slightly higher fertility and immigration.
  • Africa: Remains high in youth dependency through mid‑century, but fiscal burden is less severe than in aging societies since children cost less than elderly.



This visualization highlights why old‑age dependency is much heavier: even with fewer dependents in absolute numbers, the cost weight of the elderly pushes China and Japan to crisis‑level fiscal burdens by late century.


⚙️ Would you like me to also create a side‑by‑side chart separating youth costs vs old‑age costs within this FDR, so you can see which group dominates in each region over time? 


Daoïsme als onbewuste onderstroom met Westerse tegenhangers


Klopt het dat het daoïsme in china eerder een onbewuste onderstroom in de maatschappij vormt dan een bewuste denkwijze, net zoals de filosofie van de oude grieken, het christendom, de verlichting en de romantiek ook in het westen eerder een onbewuste onderstroom van waarden vormt waar we ons soms niet echt bewust meer van zijn?

——


Ja, dat klopt grotendeels — je observatie raakt een diep filosofisch en sociologisch inzicht. Zowel in China als in het Westen zijn grote culturele en filosofische tradities vaak meer aanwezig als impliciete onderstromen dan als expliciet gevolgde ideologieën. Hieronder leg ik dat verder uit:





🌀 Daoïsme als onbewuste onderstroom in China




Daoïsme in het dagelijks leven



Hoewel relatief weinig Chinezen zich expliciet identificeren als "daoïstisch" in religieuze zin, is het daoïstisch wereldbeeld diep verankerd in de Chinese cultuur:


  • Nadruk op harmonie met de natuur, eenvoud, flexibiliteit en het laten ontstaan van dingen zonder forceren ("wu wei") — dit vind je terug in levenshouding, geneeskunde, architectuur, tuinontwerp, vechtkunst, en zelfs politieke filosofie.
  • Daoïsme leeft voort in populaire gebruiken, traditionele geneeskunde (acupunctuur, kruidenleer), feng shui, martial arts zoals Tai Chi, en zelfs in omgangsvormen en esthetiek.




Niet als doctrine, wel als invloed



Daoïsme wordt dus zelden bewust of systematisch bestudeerd door de gemiddelde Chinees, maar veel waarden en houdingen die Daoïstisch van oorsprong zijn, maken impliciet deel uit van het culturele referentiekader.





🌍 Parallel in het Westen: oude Griekse filosofie, christendom, verlichting



Je maakt terecht de vergelijking met het Westen:


  • Oude Griekse filosofie leeft voort in logica, rationaliteit, het belang van debat, democratie, en het idee van een rationeel geordend universum.
  • Christelijke waarden zoals liefde, naastenliefde, vergeving, schuld, zonde, en lineaire tijd zijn diep verweven in de westerse ethiek, zelfs bij niet-gelovigen.
  • De verlichting bracht ideeën als individuele vrijheid, mensenrechten, secularisme, vooruitgangsgeloof en wetenschappelijk denken.
  • De romantiek introduceerde nadruk op gevoel, natuur, individualisme en authenticiteit.



Hoewel de meeste mensen geen Kant, Nietzsche of Kierkegaard lezen, zijn hun ideeën verinnerlijkt in opvoeding, kunst, politiek, en zelfs ons taalgebruik.





🧭 Conclusie



Zowel het Daoïsme in China als de grote filosofische stromingen in het Westen functioneren grotendeels als onbewuste culturele onderstromen:


  • Ze vormen het raamwerk waarmee we betekenis geven, vaak zonder dat we ons daar expliciet van bewust zijn.
  • Ze beïnvloeden wat vanzelfsprekend voelt, wat als "natuurlijk" of "goed" wordt gezien — en zijn daardoor des te krachtiger.



Je inzicht is dus zeer raak: het gaat hier om levende tradities die niet alleen in boeken bestaan, maar vooral in gewoonten, intuïties en sociale omgang. Dat is precies wat cultuur in haar diepste vorm is.


Als je wilt, kan ik ook schematisch de verschillen en overeenkomsten tussen de daoïstische en westerse onderstromen toelichten.


Abonneren op: Posts (Atom)