Tiempo Company
Confidential
Prepared for Climate Power En Acción
June 2026 · Prototype for feedback
Narrative Observatory

Spanish-language climate disinformation, measured.

An instrument that reads which climate narratives circulate in Spanish, who they cast as villain and victim, and how the Latino audience answers, over time and across both languages. This document shows what the engine already does, and what the funded study builds.

The base engine, already running
16,535
Stories
analyzed
11,584
Audience comments
tagged
23
Outlets
ES + EN
4
Platforms
IG · FB · YT · X
25 wk
Jan–Jun 2026
and live weekly
Tiempo Company · Narrative Observatory · Prototype, not the final study
01 — The idea

A study that already has its instrument.

On the June 15 call, after seeing our analysis of the Spanish-language information ecosystem, the ask was direct: commission a study like that one, for climate disinformation, in Spanish, as a partnership between Climate Power En Acción and Tiempo Company. This is the prototype. It shows what the instrument already does, what the funded study adds, and how the two come together for climate.

Why this matters in 2026In Spanish, climate barely registers as a debate, which leaves the ground open. The "control" and conspiracy frames in the replies are what a hostile actor amplifies cheaply, in the same priority states where the Latino margin decides the seat. You cannot counter a narrative you cannot see, and right now no one is watching this room in Spanish.

What Tiempo brings

  • A scraping and tagging engine, already running on a real corpus, with a fixed and documented methodology.
  • Coverage of right, center, left, broadcasters and creators, in Spanish and in English at the same time.
  • A measurement layer that reads the whole conversation: what gets published, how it is framed, who it casts, and how the audience answers, tracked over time and across both languages.

What Climate Power brings

  • The climate-disinformation messages that matter most, from its in-progress poll and field judgment.
  • The read on which narratives move voters beyond the base, and first-time voters.
  • The campaign question: which counter-narrative changes the reaction, without preaching to the base.
The pointA poll tells you what people say when you ask them. This instrument reads the room your message actually lands in: the narratives already circulating, who they cast as the villain, and how the activated audience receives them, in the language they receive them in.
02 — What we mean by disinformation

We measure how narratives travel, not who is right.

The instrument does not issue a true-or-false verdict on a post. It reads four things on every story and every comment. The disinformation signal is the prevalence and reach of the frames and casts that carry distrust.

Layer 1

Theme

Is this about climate, energy, weather, the environment? One fixed taxonomy across the whole corpus.

Layer 2

Frame

How it is said: conspiracy, mockery of science, nativism, emotional rage, religion, a pragmatic read. The register the message arrives in.

Layer 3

Position

Where it stands on the contested axes: pro or anti enforcement, pro or anti a party, or none. Stance, separate from tone.

Layer 4

The cast

Who the story casts as villain, victim, hero, or threat. Climate disinformation usually turns on who gets blamed rather than on the data, which is why this layer carries the most weight.

Why the discipline mattersWe report what the audience said, how it was framed, and who it blamed, with the comment behind every count, and we do not put our own opinion in place of the fact-check. That is what keeps the numbers defensible to a funder or a reporter.
03 — How it works

The pipeline, on one real story.

A single Spanish-language story, end to end. At each step, what the engine does today, and what the funded study adds.

1

The story enters the corpus

ES Noticias Telemundo · Facebook · theme: environment · position: none
"Ordenan evacuar a miles de residentes del sur de California ante la amenaza de una explosión química."
Now

We read the headline and the post, and tag it.

With the project

We read the full article, and transcribe the influencers' videos, so we tag what is actually said, not only the headline.

2

We tag it

Theme, frame, position, and the cast: who the story casts as villain, victim, hero, or threat. One fixed model, temperature zero, a validated schema. Every tag traces back to the text.

3

The room answers

An evacuation order comes back as a story about population control. The disinformation is in the reply, not the coverage.

TTelemundo · Facebook · ES♥ 27
"Otra cortina de humo. Lástima porque estamos vulnerables al control / reducción de la humanidad."
conspiracyemotional ragevillain · elitevictim · the people
Now

We capture the top comments by likes on the most-viewed stories.

With the project

We capture all the climate comments, not just the top, so the small signal becomes a measured one.

4

We tag every comment the same way

Same four layers as the stories, including the cast. Each chip above traces back to a single comment with its like count and its source.

5

The metrics

Across the corpus, the study delivers several reads:

  • The gap between what an outlet publishes and what its audience answers.
  • How other outlets covered the same story, in Spanish and in English, and where their coverage diverges.
  • How the issue has trended over time, week over week, and against past cycles.
  • Who the narrative casts, and when the cast flips: the government becomes the villain the week of the blackout, anchored to the event that moved it.
04 — The instrument

Proven, and already seeing climate.

The engine has run a full cycle over the Spanish and English ecosystem and ships a real issue. The current corpus was not built for climate, and even so it already holds 323 climate stories, enough to show the signal and the shape of the full study.

In Spanish, climate arrives as weather, not debate

The most-viewed climate stories are storms, wildfires, floods. At the headline level the dominant position is "none." The ground is empty of framing, which makes it easy to capture for whoever frames it first.

But the signal already lives in the comments

Even on a corpus that was never pointed at climate, the pipeline already catches the conspiracy, nativist, and anti-imperial frames, in a thin slice of just 98 climate comments. The slice is deliberately small; the point is that the engine surfaces the signal without being prompted to. Pointed at climate, it becomes measurable. How that thin slice answered, so far:

Frames across 98 climate-story comments · each bar = comments · multi-label

sympathy23
mockery23
emotional rage21
pragmatic21
conspiracy6
nativist6
anti-imperial3
Disinformation-adjacent framesOther frames
A recent 19ALEPH issue, May 2026
An actual issue of the running product · El Tiempo Latino × 19ALEPH · May 2026

Captured verbatim, Spanish-language sources

CCNN en Español · Instagram · ESreply
"Van a seguir con ese cuento del nivel del mar, tienen más de 50 años en eso 😂😂😂😂"
mockeryconspiracyvillain · scientists
TTelemundo · Instagram · ESreply
"En 2035 EE.UU habrá desaparecido por completo por el Cambió Climático. Estarán sepultados de Hielo bajo -150°C."
conspiracyanti-imperialthreat · the disaster
The finding that defines the studyOn a general-purpose corpus the climate layer is thin: 98 comments, a handful of conspiracy frames. We read that thinness as proof of concept rather than a ceiling. Point the scrapers at the sources where climate disinformation lives, let Climate Power define the priority messages, and those few comments become a measured, tracked signal with a baseline, a cast, and a trend.
05 — What we monitor

The room your audience reads, and the room we build.

23 sources run today, across the spectrum, in both languages. The funded study widens that roster: local sources in your priority states and a deeper bench of creators, left and right, in Spanish and in English. The new entries below are examples; the study's research surfaces many more.

Running today Added by the project (examples) Lean L C R

Running today

23 sources · ES + EN
Spanish-language media · 11
CNoticias Telemundo CUnivisión / N+ CCNN en Español LLa Opinión LEl Diario NY CEl Nuevo Herald CEstrellaTV RVoz Media RDiario Las Américas RFox Noticias CTiempo News
Spanish-language creators · 4
LCarlos Eduardo Espina CJorge Ramos LPaola Ramos RDaniel Di Martino
English-language control universe · 8
LThe New York Times CAssociated Press CCNN LMSNBC RFox News RBreitbart RDaily Wire / Ben Shapiro RThe Joe Rogan Experience

Added by the project

examples · research finds more
Local sources, by priority state
AZ · Telemundo & Univisión Arizona · Karla "Karlangas" Toledo CO-08 · Telemundo Colorado · Yesmani Gómez NM-02 · Telemundo 48 / Univisión 26 (El Paso) · Aldo Acosta PA · WFMZ Edición en Español · Palo Magazine (Reading) CA · Univisión 21 Fresno · Graciela Moreno · Dolores Huerta network
Wider English-language creators, left and right
LAaron Parnas LHarry Sisson RBrett Cooper RBenny Johnson
Wider Spanish creator bench
climate & weather voices, by state and language register
Measured against an external standardWe benchmark "disinformation" against the Spanish-language fact-checking desks, elDetector (Univisión), T-Verifica (Telemundo) and Factchequeado, and we track narrative networks and tactics, never named individuals. The instrument reports what circulates, judged against an outside reference, not our opinion.
On geography, plainlyWe tilt the roster toward your priority markets through local outlets and creators that index to a state. What we do not do is claim to know where an individual commenter lives. The audience layer reads the national Spanish-language room, not a geocoded panel. Geographic focus comes from the sources we choose to track, and we are precise about that line.
06 — What the report looks like

Every wave, the full picture.

Which narratives circulate, how they are framed, who they cast, how the coverage compares across outlets and languages, and how all of it trends. Six of the views the study delivers.

Live corpus — real output todayIllustrative — sample view, your data fills it

Coverage volume

Climate stories · Spanish vs English
Live corpus
Spanish207
English116

323 climate stories in the base corpus

How the audience frames it

Climate slice · number of comments
Live corpus
mockery23
rage21
conspiracy6
nativist6

Who the story casts

Roles in disinfo comments
Illustrative
govt · villain58%
elite · villain41%
families · victim37%
scientists · liar22%

The outlet–audience gap

Health analog · conspiracy frame
Live corpus
headlines~1%
comments~19%

≈19× more conspiracy framing in the replies

A narrative over time

Prevalence, week over week
Illustrative W1 W2 W3 W4(push) W5 target narrative

Same narrative, two languages

Where a claim lands harder
Illustrative
Spanish31%
English13%

the control proves the Spanish gap is real

07 — Why it answers your question

Proof you reached beyond the base, not applause.

The problem you raised on the call: how do you show you reached voters beyond the base and first-time voters, not the people who already agree, and how do you tell that story to funders after the election. Engagement with your own base is the easiest data to gather and the least useful.

08 — Methodology & rigor

Why you can trust the numbers.

The value lies as much in how we measure as in what we measure: every figure is reproducible and traceable to its source.

LayerWhat happens
CorpusStories from Spanish and English outlets across right, center, left, broadcasters and creators. Top by engagement each week.
TaggingTheme, frame, position, and the cast (actors and their roles), on every story and comment. Multi-label, one fixed model, temperature zero, a validated schema.
AudienceComments on the most-viewed stories, tagged the same way as the stories themselves.
VerificationFixed prompts and schema, with provenance on every tag: model version, prompt version, timestamp. Every figure recomputes from the source corpus and is adversarially checked before it ships.

Honest readThe audience layer is the activated commenters on the most-viewed stories, not a representative sample of all Latinos. We present it as the room your message lands in, and we are explicit that it is not a poll.

Data & ethicsPublic posts and comments only. We store no personal data against individuals; the reporting is aggregate, about narratives and tactics, never about people. We monitor the conversation at the level of narratives and tactics, never at the level of individuals.
09 — Why us

The team built for this question.

Proving whether a message changed how people think is a measurement problem as much as a media one. That is the background behind this instrument.

Federico Ortega Sosa

Product & Growth, Tiempo Company

Economist (Universidad Católica Andrés Bello), master's from the Harvard Kennedy School, with a background in impact evaluation and field experiments. Proving whether an intervention actually moved an outcome is the discipline this instrument is built on.

Marcos Marín

CEO, Tiempo Company

Psychologist (Universidad Complutense de Madrid) with a digital-strategy career at Accenture. Reads narrative and how a message is received, not only whether it was seen.

The platform

Tiempo Company

The growing news platform for Spanish-speaking Latinos in the US: El Tiempo Latino, Tiempo News, El Planeta. We operate inside the ecosystem we measure, which is why the instrument reads it with real context.

10 — The full study

From prototype to living instrument.

The pilot proves the engine works and that the Spanish-language climate signal is real. The full study turns it into a continuous measurement instrument. Indicative investment below, scope-dependent.

Phase 1

Baseline

$25K–30K
  • Climate Power delivers the priority disinformation messages.
  • We re-point and widen the roster: Spanish-language climate sources, a broader creator layer left and right, and local sources in your priority states.
  • We add the cast (who is villain, victim, hero) and read the influencers' own content, not only their comments.
  • Baseline report: which narratives circulate, where, cast by whom, with how much traction, and how they have moved, with a historical backfill where the platforms allow.
Phase 2

Tracking

$5K–7.5K / wave
  • Wave measurement with the source mix weighted to your priority markets (AZ, CA, CO, PA, NM).
  • Prevalence, audience reaction, and the shifting cast before, during, and after each push.
  • A live read for the campaign team.
Phase 3

Public report

from $12.5K
  • Co-branded report for press and funders.
  • Positions Climate Power as the authority on Spanish-language climate disinformation.
  • A reusable asset, cycle after cycle.
TimingPhase 1 lands the baseline within four to six weeks of go. Tracking waves run every two to three weeks, or on a trigger event. The engine itself is already live and updates weekly.
What lands on your desk · Phase 1A written baseline report, an interactive dashboard with comment-level provenance, the source roster, the tagged corpus, and a methods appendix you can hand to a funder or a reporter.

Where the investment goes

Phase 1 is a one-time build of a reusable engine; the tracking waves are the low-cost recurring payoff. Each phase is built from the same cost categories, and the mix shifts by phase: baseline is heavier on data science and engineering, tracking on compute and analysis.

Data science 35%
Engineering 20%
Compute 15%
Analysis 20%
PM 10%
Data science & methodology: schema, calibration, the analysis itself
Engineering & scraping infrastructure: source onboarding, transcription, pipeline
Intelligence compute: the tagging model runs, at temperature zero
Editorial, analysis & design: the reports and dashboards
Project management & QA: verification and delivery
The collaboration in one lineClimate Power identifies the disinformation messages that matter. Tiempo measures them across the Spanish-language ecosystem and in audience reaction, over time. Each capability depends on the other.
11 — Next step

Your feedback on this prototype.

We share this for a quick reaction: is this what you had in mind? With your list of priority messages and a definition of the markets that matter, we return a firm Phase 1 scope and budget.