A 360° real-time read on the political attention layer six months before the 2026 midterms — what narratives are driving influence, who is gaining and losing momentum, and what four frontier AI models conclude when fed the same dataset.
Trump's 1.03B daily engagements equal roughly 40 times the entire combined volume of all named Democratic figures.
Of 4 themes detected in MAGA-tagged conversation, 3 are critical:
All 4 LLMs converged on this as the single most important data point in the entire dataset.
Daily engagement volume across 12 named political figures. Bars normalized to Trump's volume. Sentiment computed at the post level then aggregated across 24h.
Daily engagement curves over the last 6 months for the four most diagnostic political topics. Trump's volatility dwarfs everyone else; DeSantis's late-April liftoff is visible; Newsom's collapse is unmistakable.
GOP gov nominee, Ohio · won primary May 7
52-week highs across the board, today. Trump-endorsed primary win. 3,334 unique creators — also 52-week high.
FL Governor · presumptive 2028 GOP candidate
Structural momentum, not a news spike. +83% creators, +105% mentions, +107% posts — all 30-day deltas. Hit 52-wk highs May 2.
Senator (I-VT) · progressive figurehead
Strongest left-flank energy in dataset. 77% sentiment is the highest of any Dem-aligned figure. Israel-aid wedge gives cross-coalition reach.
CA Governor · presumed 2028 Dem hopeful
Outright collapse, not slowdown. Volume down ~10× from Nov peak. 70% critical on policy/economy + 30% on $267M hospice fraud scheme.
Former VP · 2028 positioning
Below-the-line continuation. Critical themes (30%) outweigh supportive (25%). 2024 campaign criticism still dominant frame.
Conservative commentator
Once a unifier, now a wedge. 25% antisemitism accusations + 15% foreign-agent allegations + 10% Trump break — his audience is fracturing.
The number is loud, but the methodology behind it makes it interesting. "15% supportive" doesn't mean only 15% of people support MAGA. It means something more specific — and more diagnostic of the 2026 cycle.
Tone of MAGA-tagged posts in the last 24 hours. Positive-leaning.
Sentiment measures how nicely people are speaking. It says nothing about what they're speaking about.
Themes detected by NLP across the same posts. Three are critical.
Themes measure what the conversation is about. The mix has flipped.
The MAGA-positive crowd is going quiet — and the people still talking about MAGA are talking about disagreements within it, not about why it's winning.
The 60% theme is about Trump himself: not about Dems, not about media, not about RINOs. The biggest critical conversation inside MAGA is about Trump's recent decisions.
Midterm GOP overperformance has historically required base mobilization, not swing voters.
Enthusiasm shows up as supportive-theme volume. Today MAGA's own topic produces 1 supportive theme to 3 critical — the inverse of the 2024 cycle. If the mix is still 1:3 in October, GOP underperforms its structural floor by 5–15 House seats and likely loses the Senate.
Trump-backed candidates won 26 of 26 Republican primaries. But internal MAGA divisions are growing.
The electoral machine is still working. The cultural movement underneath it is fracturing. Those usually move together — when they don't, the machine eventually breaks. That's the variable to watch through October.
95% of MAGA's current themes are critical of Trump's actions or the movement itself.
Only 15% are supportive.
This is not external opposition. This is conversation tagged with the President's own movement turning critical of its leader. Four AI models analyzing this dataset independently agreed: it is the most important data point in the entire 2026 cycle.
The same LunarCrush dataset and prompt sent to four frontier AI models. Their responses converged with materially identical reasoning. Cross-family consensus on a political prediction is itself an unusually strong signal.
DeSantis · Vivek Ramaswamy · Bernie Sanders
Same three figures named by every model with materially identical reasoning.
Newsom · Harris · Tucker Carlson
Same three figures named by every model with same evidence chains.
MAGA's 95% critical-theme weight
Four labs, one number. The cover-of-the-briefing data point.
If MAGA cohesion holds through October, GOP overperforms. If it fragments further, statewide races flip. Confidence: medium. All four LLMs converged.
Structural floor is too high to flip.
Statewide races are more variance-sensitive.
Concrete metrics whose movement will signal cycle direction before traditional polling registers it.
Currently 61%. Cross below 50% = cycle has measurably turned.
Cycle-direction triggerCurrently 46%. Recovery above 55% = right-amp stack reunifying. Decline below 40% = active fracture continuing.
Cohesion proxy+83-107% over 30 days. Creators-per-day above 5,000 through summer = he becomes dominant non-Trump GOP brand.
2028 invisible primaryCurrently 52-wk low (134 mentions/day). Recovery above 2,000/day = scandal contained. Continued decline = 2028 candidacy nonviable.
Dem successor question+39% posts WoW. Sustained through Q3 = progressive lane consolidated. Watch for him to anoint a successor — high-leverage 2028 signal.
Successor signalTrump's Iran themes 35% supportive / 20% critical today. If critical surpasses supportive, the President's #1 narrative driver flips from asset to liability within days.
Foreign-policy betaReal-time social intelligence at platform scale, normalized into structured signal that a decision-maker can read in minutes.
30M-50M posts per hour across X, Reddit, TikTok, YouTube, Instagram, plus 10,000+ news sources. Continuous coverage of every named entity.
Per-topic time-series with platform breakdowns. Sentiment computed at the post level using LunarCrush's classifier.
For each topic, dominant supportive and critical themes are extracted with weighting (% of conversation). Reveals what people are saying, not just how loudly.
Per-topic ranking by amplification. Reveals which voices actually move each conversation, not just who has the most followers.
Full structured dataset sent to four frontier AI models from four labs with the same prompt. Convergence functions as a confidence signal.
This dashboard. Every claim traces to a specific data point. No external polling. No prior-knowledge pattern matching. Pure social-attention signal interpreted by AI.