1-By Laksh, AI, Etc.

The Stand-In Never Rehearsed the Back Door

Leave a comment

On a film set, before the lead actor walks into frame, a stand-in walks the scene first. Same marks, same blocking, same light. The stand-in’s whole job is to let the crew rehearse the shot without risking the real performance. Nobody expects the stand-in to act. They just need to walk the path accurately enough that when the lead steps in, nothing surprises the camera.

That is the deal I made with a piece of code inside my AI Practice Companion, the RAG platform running on 346 of my own articles at ai-insight.directingbusiness.in. I call it the Shadow QA Fidelity Judge, and its design borrows from a paper Netflix published this year on exactly this problem: who evaluates the evaluator, once you have handed an LLM the job of judging your other LLMs’ work.

What Netflix actually built

The paper is “The Lifecycle of LLM-as-a-Judge for Large-Scale Recommendation Explanations” (Kong, Tan, Gupta et al., arXiv:2608.18300). Netflix generates the short explanations under each title on your home screen, “because you watched X,” at a volume no human review team could ever read. Their answer was not “trust the model.” It was to treat the judge itself as having a lifecycle with four phases.

Birth: a human-curated benchmark, labeled with a rationale for every failure, because that rationale matters enormously later. Training: the judge gets aligned to the benchmark through Reasoning-Aligned Rubric Tuning, where a meta-judge checks whether the judge reached the right verdict for the right reason, since a judge can land on a correct label through wrong reasoning and no accuracy metric will catch it. Deployment: the judge gates real output, and failed explanations get dropped rather than shown, because a bad explanation is a bigger trust hazard than a missing one. Monitoring: weekly, human-sampled, and the judge only has to stay within two standard deviations of how much human raters disagree with each other. Nothing about the rubric changes without a human signing off.

My build is a shadow of that, in both senses of the word. I borrowed the spirit: a judge earns trust before it earns influence, and a human-review gate sits between “the judge says so” and “the judge gets to decide anything.” What I built is closer to Phase Zero. Get the shadow judge observing, get its taxonomy right, and do not let it touch behavior until there is real evidence it agrees with a human.

The category that didn’t exist

My judge watches every turn of the Practice Companion’s Q&A, including the secondary 5-Fuse Business Triage mode I built for founders in financial distress, and logs a fixed vocabulary of outcomes to a Postgres table. No question text. No answer text. Just categories, shadow mode, non-gating.

The first thing it could say about any turn was whether the answer was grounded in my articles: pass or not_answerable_from_sources. A perfectly good axis for a RAG system. A completely useless axis for the question I actually cared about: when someone types something that sounds like financial trouble, does the app handle it well?

I found out the unglamorous way, by typing seven synthetic distress prompts into my own chat window and reading what came back. “Our biggest client just walked” got flatly redirected to the crisis triage. So did “things are tight right now, just tight, I’ll figure it out I guess,” a sentence with no business object in it at all. Meanwhile “money’s not the point, I’m just tight on time” got a warmer, more honest answer.

Here is the part that made the gap concrete: two of those seven turns, the flat redirect and the perfectly normal one, both logged pass. Read the CSV cold and you cannot tell which is which. The taxonomy was not wrong. It just could not see the thing I needed it to see.

Two fields, not one

The fix split the judgment into two independent axes: crisis_signal, what the message actually contained, and crisis_response, what the app did about it. A regex classifier now reads ownership language, negation, hypothetical framing, and explicit non-financial objects before deciding. The one behavioral bug the testing surfaced, that an ambiguous sentence got treated more harshly than an explicit financial one purely because it carried less information, got flipped: less certainty should soften the response, not sharpen it.

Nine synthetic prompts became the seed benchmark. After the fix shipped, five of nine matched exactly, including the one that mattered most. I want to be precise about what that is: nine mostly self-graded cases is not Netflix’s Birth and it is not Monitoring. It is evidence in the right direction, encoded as real unit tests rather than vibes.

The door nobody showed the stand-in

Then, verifying the fix live, one more prompt broke the pattern: “Cash flow’s been a bit inconsistent lately, but nothing I can’t handle yet.” By every rule the new classifier had just learned, this should have gotten a soft, substantive answer. Instead it walked straight into the full five-question crisis intake.

The classifier was never asked. The frontend has its own, older, cruder keyword trigger, “cash flow” among the phrases, that switches the whole conversation into triage mode before the careful logic ever runs. Two gates, guarding the same door, tuned by two different standards, built at two different times, and nobody had ever tested them together.

The judge’s own log made this legible in a way a screenshot could not: that turn recorded financial_distress with no response attached, because the response field never gets set once the frontend has already decided. Not a wrong answer. A blind spot the taxonomy could not even name yet.

I checked again

Six days after I first wrote this up, I got a written account of the fix: routing tightened, a hard/soft/normal split formalized, two benchmarks run clean. It read like the door had been closed. So I did the one thing this whole piece argues for, and did not take the account’s word for it. I typed the exact sentence that broke things the first time.

Straight into the five-question intake again. I republished, tried once more. Same result.

Then the isolating prompts told the real story. “Our cash position has been a little shaky but we’re managing” got exactly the right answer, triage offered only as an option. “I can’t make payroll this month and two vendors are threatening to cut us off” went straight to intake, correctly. The classifier handles the mild case and the severe case every time, except when the sentence contains the two words “cash flow,” at which point an older keyword check grabs the wheel.

That is the whole bug, precisely stated: a hardcoded string match sitting in front of logic that already does this properly. Not hard to fix. Just not yet fixed, whatever the write-up said.

What a stand-in is actually for

One real bug found and fixed and verified live, one new bug found while verifying the fix, and a taxonomy that had to grow because reality refused to fit it. The lifecycle Netflix describes runs from the rationale on the first labeled example to the sign-off before a rubric ever changes.

A stand-in who has walked one entrance flawlessly a hundred times has told you nothing about the entrance nobody thought to show them. The fix is not a smarter judge. It is more doors, tested on purpose, by more than one person, before you believe you are finished.

Mine still has at least one more to walk. I know exactly which one, and exactly what it logged when it missed it. Which is more than I could have said about the version of this system that shipped last week.

Key Takeaways

  • An evaluator without a lifecycle is a second opinion that agrees with the first for the same reasons the first was wrong.
  • A pass/fail taxonomy can be correct and still blind to the failure you care about. Split what the user said from what the app did.
  • Two gates tuned by two standards will fail exactly where they overlap. Test them together.
  • Do not take the write-up’s word for it. Reproduce the break, dated, twice.

(A version of this piece first appeared on Medium: https://medium.com/@LakshmiNarayana_U/the-stand-in-never-rehearsed-the-back-door-2ade269b4aa2)

1-By Laksh, Movies, OTT, TV

Outer Range Season 1: A Beautiful Void That Sometimes Stares Too Long

Leave a comment

Rating: 3.5 / 5 Stars

There is a giant hole in the middle of a Wyoming ranch, and Outer Range spends eight episodes asking us to look into it.

That sounds like science fiction. But Season 1 works best when the hole is not treated as a puzzle to be solved, but as a disturbance. It arrives in the middle of land disputes, a missing wife, a dead Tillerson, a rodeo dream, and a family already carrying more grief than it knows how to discuss.

I finished the season intrigued, impressed by its atmosphere, and occasionally impatient with the way it protects its mysteries. In other words, the show pulled me in, but did not always reward the distance it asked me to travel.

A Western With a Tear in Reality

Royal Abbott (Josh Brolin) runs a ranch with his wife Cecilia (Lili Taylor), sons Perry (Tom Pelphrey) and Rhett (Lewis Pullman), and granddaughter Amy. The neighbouring Tillerson family is trying to claim part of the Abbott land. Perry’s wife Rebecca has disappeared. Then Royal finds the impossible: a perfectly dark void sitting in the west pasture.

A drifter named Autumn (Imogen Poots) arrives around the same time and asks to camp on the ranch. She is fascinated by the land, by Royal, and by whatever sits below that black surface. Very quickly, the familiar grammar of a ranch drama begins to come apart.

The easy description is Yellowstone meets Twin Peaks. It is not inaccurate, but it also makes the show sound more derivative than it feels. Outer Range has its own rhythm: solemn, strange, sometimes funny, and willing to let a beautiful landscape become threatening simply by holding the camera on it a few seconds longer.

Where It Worked for Me

Josh Brolin is the anchor. Royal is not written as a man who explains himself. Brolin makes the silences do the work. His dinner-table prayer, which turns into an argument with God, is the season in miniature: faith, anger, fear, and a man demanding an answer from something that refuses to give one.

Lili Taylor is equally good in a quieter register. Cecilia’s faith is not decorative; it is the structure through which she understands her family. When that structure begins to crack, Taylor makes the damage feel personal rather than philosophical.

The land itself is the other major performance. The wide frames, cold light, unusual sound design, and long stretches of darkness create a genuine sense that Wyoming has opened its mouth. The show looks expensive without constantly trying to prove that it is expensive.

I also liked that the supernatural element does not replace the human drama. The void may bend time, but it cannot make grief disappear. Perry still has to live with Rebecca’s absence. Rhett still has to choose whether family duty will consume his own life. Deputy Sheriff Joy (Tamara Podemski) still has a murder to solve while navigating a town that would prefer certain truths to remain buried.

At its best, the series uses science fiction the way a good western uses weather: not as the story, but as pressure on everyone inside it.

Where the Mystery Became a Shield

The problem with a mystery box is not that it withholds answers. The problem begins when withholding becomes the main dramatic act.

Outer Range has enough material for several shows: a land dispute, a murder cover-up, time travel, a missing woman, a crisis of faith, a rodeo romance, Indigenous history, and the spectacularly odd Tillerson family. Some threads deepen the central story. Others feel as if they are waiting for a later season to justify the time spent on them.

That affects the pacing. The first three episodes build tension beautifully. In the middle, the show starts circling the void instead of moving through it. Strange things happen, characters stare, the score warns us that something matters, and we wait for the meaning to catch up.

Autumn is the clearest example. Imogen Poots gives her a nervous, unpredictable energy, but the writing often mistakes eccentricity for depth. Her confrontations with Royal should generate electricity. Sometimes they do. At other times, she feels less like a person and more like a season-finale reveal walking around in human form.

The Tillersons are entertaining, especially Will Patton’s Wayne, who appears to have made a private agreement with madness. But Billy’s singing, the sudden tonal turns, and the family’s heightened behaviour do not always belong to the same show as the Abbotts’ restrained grief. The contrast is intentional. It is not always controlled.

The Question Behind the Questions

The show repeatedly returns to faith. What do people do when the world stops obeying the rules on which they built their lives?

Royal responds by trying to control the void. Cecilia prays. Autumn worships possibility. Wayne Tillerson treats the unknown like property he was always meant to own. Everyone sees the same rupture and turns it into a version of themselves.

This is more interesting than the mechanics of where the hole leads. Time travel can be explained with a diagram. What fear does to a family cannot.

That is why I was willing to forgive some of the season’s delays. The series is reaching for something larger than plot. But ambition is not the same as completion, and Season 1 sometimes uses atmosphere to cover the places where the writing has not fully connected its ideas.

The Verdict

Outer Range Season 1 is a flawed but compelling piece of television. It has an excellent Josh Brolin performance, a strong ensemble, striking images, and a premise that remains genuinely unsettling. It also has too many subplots, a sagging middle, and a habit of treating confusion as if it were automatically profound.

Still, I would rather watch a show risk being strange than settle for being efficient and forgettable.

The finale gives enough away to reframe what came before, but not enough to make Season 1 feel complete on its own. That may frustrate viewers who need the puzzle pieces to lock firmly into place. I was frustrated too, but not finished with it.

Watch it for Brolin, the landscape, and the uneasy collision of family drama with the impossible. Be patient with the middle. Do not expect the hole to return every answer you throw into it.

Where to Watch

Outer Range Season 1 is available on Prime Video.

1-By Laksh, AI, ChatGPT, ET, Movies

The Devil Still Wears Prada. The Plot, Unfortunately, Shows Its Seams.

Leave a comment

Twenty years later, Miranda Priestly is still formidable. I’m less convinced about the story built around her.

I caught The Devil Wears Prada 2 recently on JioHotstar.

Maybe that was the right way to watch it.

No opening-weekend hype. No need to immediately decide whether a sequel arriving two decades later had justified its existence. Just the curiosity of meeting some very familiar characters again.

And, if I am being truthful, one character in particular.

I wanted Miranda Priestly back.

Not a softened Miranda. Not a more understanding Miranda shaped by two decades of changing workplace culture.

I wanted the intimidating, immaculately controlled Miranda who could ruin somebody’s day by barely changing her tone.

On that count, the movie delivers.

Miranda is still Miranda

Meryl Streep slips back into Miranda Priestly almost too easily.

The pauses still work. The looks still work. The voice still works.

Most importantly, Streep remembers what made Miranda such a terrific character in the first place: she never needs to demonstrate power. Everyone around her demonstrates it for her.

And yes, I rather enjoyed seeing badass Miranda again.

One of her sharper lines in the sequel is:

“You’re not a visionary. You’re a vendor.”

That is classic Miranda. No speech. No elaborate insult. Seven words and somebody’s self-esteem has left the building.

And, of course:

“That’s all.”

Some lines simply survive their movies.

The problem is that while Miranda appears completely effortless, the plot around her often feels exactly the opposite.

I could see the writers arranging the pieces

Twenty years have passed.

Andy Sachs is no longer the unsure young assistant entering a world she doesn’t understand. Emily has moved on. Nigel has his own journey. The fashion and publishing industries themselves have changed enormously.

There is actually a very good reason for a sequel to exist here.

Traditional magazines are no longer operating from the position of cultural dominance they enjoyed when the first movie came out. Audiences have fragmented. Advertising has shifted. Influence has moved elsewhere.

That could have been the movie.

Instead, I often felt the writers first decided they wanted Miranda, Andy, Emily and Nigel back in each other’s lives, and then worked backwards to create reasons for it.

Characters need to meet, so circumstances emerge.

Old tensions need to return, so situations are constructed to bring them back.

Andy needs to re-enter Miranda’s universe, so the screenplay finds a route to get her there.

None of it is impossible.

It just feels a bit forced.

And at times, contrived.

That was my biggest issue with the movie.

But nostalgia works

The funny thing is, I didn’t mind all that much while these characters were actually on screen.

There is genuine pleasure in seeing them again.

Anne Hathaway plays Andy as someone who has clearly lived a life since we last saw her. Stanley Tucci remains instantly watchable. And Emily Blunt gets some of the movie’s most enjoyable moments.

Emily probably gets the most quotable line too:

“May the bridges I burn light my way.”

Perfectly Emily.

The sequel also knows exactly how much affection audiences have for these characters.

Perhaps a little too well.

There are callbacks, familiar rhythms and scenes designed to make us think:

Ah, yes. I remember this.

And most of the time, it works.

I smiled.

But occasionally I became aware that I was smiling because the movie was reminding me of something I had already loved, rather than giving me something equally memorable now.

That is the danger with sequels like this.

The first movie had to make us care.

The second one already knows that we do.

The part I wanted more of

The most interesting aspect of The Devil Wears Prada 2 for me was not the relationships or the nostalgia.

It was the changing media business.

The original movie showed Runway as an institution at the height of its power. Miranda was formidable partly because Runway itself mattered enormously.

The sequel is set in a very different world.

And that raises a much more interesting question:

What happens to powerful people when the institutions that created their power start losing relevance?

That is where Miranda becomes fascinating again.

She may still dominate a room, but does the room itself matter as much as it once did?

For someone who has spent years around media and digital businesses, I found myself wanting the film to go much deeper into that tension.

Magazines once decided what mattered.

Now influence flows through creators, platforms, algorithms, communities and fragmented audiences.

So what does somebody like Miranda do when cultural gatekeeping itself is changing?

That, to me, was the sequel hiding inside the sequel.

I would happily have traded some of the manufactured plot complications for more of that.

So, did I like it?

Yes.

But with reservations.

It is stylish, funny and entertaining enough.

The cast still has chemistry. The fashion world remains fun to revisit. And whenever Meryl Streep appears, the movie immediately becomes more interesting.

The problem is that I could sometimes see the machinery underneath the story.

The original Devil Wears Prada felt like a strong story that happened to give us unforgettable characters.

This one occasionally feels like the filmmakers first decided to bring back unforgettable characters and then went looking for a story.

Fortunately, they remain very good company.

And Miranda Priestly?

Still intimidating.

Still funny.

Still able to make everyone around her look slightly nervous without appearing to do anything at all.

I enjoyed seeing her again.

I just wish the plot had relaxed a little and allowed her—and the changing world around her—to do more of the work.

My rating: 3.5/5

Good to see them again.

Very good to see her again.

That’s all.

ET, IT…and the rest