Gemini analysis of my taste in movies

(Gemini) MovieLens.org is a non-commercial movie recommendation engine and research platform operated by the GroupLens Research lab at the University of Minnesota. Since its launch in 1997, it has served a dual purpose: providing users with high-accuracy personalized film suggestions and serving as a critical “living laboratory” for the study of collaborative filtering and human-computer interaction.

Unlike commercial alternatives like IMDb or Letterboxd, MovieLens is entirely free of advertisements and corporate sponsorship. Its primary output is academic insight. The site is famous for pioneering early collaborative filtering algorithms which predict how you will rate a film based on your historical behavior and the behavior of similar users.

Last year I spent a few hours on MovieLens.org rating 435 movies I had watched. I think my criteria was “would I watch this movie again?” I uploaded those those movies and my ratings and asked Gemini for an analysis.


Based on your 435 ratings, I’ve performed a skeptical audit of your cinematic preferences. I’ve bypassed the “predictive recommendations” fluff to look at where your taste actually diverges from the consensus.

1. The “Lebowski” Anomaly (Contrarian Profile) -You are significantly more critical than the average MovieLens user. While the site average is ~3.72, your mean rating is 3.41. You are most “contrarian” when it comes to established cult classics and high-concept stylistic films.

  • The Big Lebowski (1998): You gave this a 1.0, while the average is 3.99. This is a massive 3-point delta.
  • The Hateful Eight (2015): You gave it a 1.0 (Avg: 3.89).
  • The Abyss (1989): You gave it a 1.0 (Avg: 3.68).
  • Scott Pilgrim vs. the World (2010): You gave it a 1.5 (Avg: 3.77).

Analysis: You appear to have a low tolerance for “quirk” or indulgent stylistic flourishes that others label as “genius.”

2. The “Maverick” Picks (Guilty Pleasures or Hidden Gems?) – Conversely, you value certain films far higher than the general public. Your “Maverick” score is highest for:

  • Duets (2000): You gave it a 5.0 (Avg: 2.85). This is your largest positive divergence (+2.15).
  • World War Z (2013): You gave it a 5.0 (Avg: 3.30).
  • The Bourne Legacy (2012): You gave it a 5.0 (Avg: 3.43).
  • Mosul (2019): You gave it a 5.0 (Avg: 3.52).

Analysis: You seem to favor competence and high-stakes procedural/action narratives (Mosul, Bourne, Sicario) over “vibe-based” cinema.

3. Historical Bias (Decade Analysis) – Despite your skeptical leanings, you have a soft spot for the mid-century.

  • 1950s: Your highest-rated decade (Avg: 3.80).
  • 2010s: Your most critical decade (Avg: 3.30).
  • 1990s: Where you spent the most time (100 movies rated).

4. Interest Alignment: Radio & Mechanics – Given your interest in radio history and mechanical vehicles:

  • Radio: You gave Broadcast News (1987) a perfect 5.0. You seem to appreciate the “transmission” of truth or the mechanics of the medium.
  • Mechanical/Procedural: You rated Sicario (2015) a 5.0, which aligns with your preference for technical precision and unsentimental realism.
  • Consciousness/Mind: You gave A Beautiful Mind (2001) a 2.0, suggesting a skepticism toward Hollywood’s “glossy” portrayal of cognitive struggles.

You are a Pragmatic Realist. You reward movies that execute a specific function well (thrillers, procedurals, mid-century classics) and punitively down-rate movies that rely on irony, surrealism, or “cult” status to earn their keep.


Hmm. Not sure I agree but I guess the data doesn’t lie.

Watching movies/TV on the iPhone

The remote for our big LG TV died so I watched an episode of Mayor of Easttown on my iPhone last night. I do that from time to time and once again I found this a very satisfying experience. At no time did I have a sense of “this is such a tiny little screen.”

I asked Perplexity about typical movie/TV viewing distance on mobil phones:

“Studies that look at smartphone viewing distances for video and general “digital device” use show that phones are commonly viewed at around about 12–14 inches, similar to reading distance. Experimental work that tested video viewing at 20 cm vs. 40 cm used those as realistic “near” and “far” smartphone distances, supporting the idea that typical use falls somewhere between them, with averages in the low 30‑cm range.”

That’s pretty much me. Had the phone propped against a pillow on my stomach. This works for me (and others?) because the relative screen size is about the same, as I attempted to show in this 10 second video. Continue reading

One Battle After Another

I rarely recommend a movie because no one ever really knows what someone else is going to like, no matter how well you know that person. Having said that…

Last night we paid $20 to rent One Battle After Another, starring Leonardo DiCaprio, Sean Penn, Benicio Del Toro and directed by Paul Thomas Anderson. The film ran 2 hours and 45 minutes but the pacing was so fast –from the opening scene– it seemed much shorter.

Sean Penn gave an amazing performance as Col. Steve Lockjaw, the film’s bad guy and if you watch closely you’ll see a few scenes with Wood Harris (Avon Barksdale from The Wire). Some of the humor and the soundtrack reminded me of Quentin Tarantino but I’m not sure Mr. Tarantino could have come up with the Christmas Adventurers Club.

One Battle has earned more than $200 million and most reviews have been glowing (“A near-perfect blend of action, humor and political urgency.”)

How will drones change future war movies?

While the Ukraine Russian war is still going on, it’s obvious the use of drone technology is changing that war and probably all future wars. I’m wondering how this technology is going to translate into the films that are made about this era.
ChatGPT: This gets at how storytelling evolves alongside military technology. Just as trench warfare shaped WWI films, and helicopters became the icon of Vietnam movies, drones will inevitably define how 21st-century wars are remembered on screen. A few thoughts on how filmmakers might handle it:

Continue reading

The Ten Best U.S.-War Films

Let’s see your list of the 10 best war movies ever made. And to make an impossible task a little less impossible, let’s limit it to wars in which the United States has been involved.

ChatGPT: Since we’re narrowing it to wars in which the U.S. was directly involved, I’ll pull across a mix of classics, modern interpretations, and films that capture both the brutality and psychology of war. Not necessarily the definitive top 10, but ten that stand up well as “best ever.” (in no particular order) Continue reading

Why is the coffee cup always empty?

I can’t remember ever having seen a scene in a TV or a movie where one of the characters is drinking from a coffee cup that it isn’t completely obvious that that cup is empty wondering why the producers directors the people making this series or movie wouldn’t wanna have something in the cup  Once again, ChatGPT has the answer.


ChatGPT: That’s one of those little things that, once you notice it, you can’t unsee it — the infamous empty coffee cup acting. There are a few reasons it happens, even though it definitely breaks the illusion: Continue reading

MovieLens.org

Editor’s Note: In the two years I’ve been playing with AI I’ve shared what I thought were some of the more interesting “conversations”, always being careful to distinguish something I wrote from AI output. ChatGPT frequently offers to compose a blog post and the drafts are as well written (better?) than mine. But they’re not mine. ChatGPT wrote (or plagiarized ) the following.


MovieLens is a free, non-commercial movie recommendation website developed by the GroupLens research lab at the University of Minnesota. It’s designed to help users discover movies they’re likely to enjoy based on their individual tastes. The site uses collaborative filtering algorithms to generate personalized recommendations, improving as more ratings are added.

Getting started is simple: users rate movies they’ve already seen using a five-star scale. The more films you rate, the better the system becomes at understanding your preferences. MovieLens then offers tailored suggestions that can be filtered by genre, release decade, popularity, and more.

One of the key strengths of MovieLens is its clean, ad-free interface and its emphasis on transparency. It doesn’t promote content for profit, and it doesn’t track users for advertising purposes. The recommendations are driven purely by user data and academic research.

Users can view their own rating history, create custom movie lists, and explore trends or themes in their viewing habits. While it lacks the social features of some commercial services, its core focus is accuracy and simplicity in recommendation.

MovieLens is especially useful for cinephiles who want to keep a record of what they’ve seen and find new films aligned with their personal taste, without the influence of box office hype or crowd-sourced noise. It’s also a valuable tool for researchers and students studying recommendation systems, as its algorithm and structure have been used in numerous academic projects and publications.