On Election Day 2026, 1.6M posts from 724K authors discussed the US midterms. Just over a fifth came from outside the United States, and those posts were about different things. This page breaks the night down by theme, by person and by where the poster was.
Across 27 hours on Election Day 2026, 1.6M posts from 724K authors discussed the US midterms, peaking at 119K posts in the 11PM ET hour as control of the House was called. 25% came from outside the United States, across 192 countries. Americans argued about turnout and fraud; the rest of the world talked about Iran, tariffs and technology. The angriest subject, election integrity & misinformation, was also among the least shared, and Donald Trump, who was not on any ballot, appeared in 23% of everything collected.
Figures may be quoted with attribution to Quid. They measure online conversation, not voting intention and not opinion in the electorate.
Volume climbed all evening and broke in the 11PM ET · 8PM PT hour. That single hour carried 7.6% of the whole window, 8.6 times the quietest hour's 14K.
Tariffs & trade drew 5.18 engagements per post; Education & schools drew 0.12. Election integrity filled 41K posts and earned 19K engagements: loud, and largely ignored.
Net sentiment of -81.9 on 2.7K labeled mentions, the most negative of the 16 themes measured. Results & race calls was the warmest at +44.9.
Posts from outside the US ran +14.0 net sentiment against +7.3 inside it. Distance takes the edge off: the angriest themes are also the most American ones.
Election integrity & misinformation was 92.2% American. AI & technology was the most international at 60.1%. The further a theme sits from the ballot, the further it travels.
Hover or tap any hour to see what was actually being posted in it, inside the US and outside, with links to the original posts.
Buckets are pinned to US Eastern, which is how poll closings are reported, and labelled in both Eastern and Pacific in the hover. Twitter/X volume is projected from a 10% sample, here as everywhere on this page.
Sixteen themes, eleven appearing in the campaign-week report. A post can belong to more than one theme, so the columns do not sum to the total.
Bars show the positive/negative split of labeled mentions only. The net score is suppressed and marked thin below 250 labeled mentions, because a split on a handful of posts is noise.
Engagements are likes, comments, shares and views. The ordering differs sharply from the volume ranking.
Americans watched a process story; everyone else watched a consequences story. Inside the United States the conversation was about the machinery of voting: election integrity & misinformation ran 92.2% American and turnout & voting access 88.9%. Those are arguments about whether the count can be trusted, and they barely crossed a border.
Outside the United States the same event was read for what it would change. Relative to its own conversation, the rest of the world over-indexed hardest on foreign policy & iran (+2.0 points against the US share) and ai & technology (+0.8) — a foreign war, trade and technology policy, not ballots. The mirror image holds: the US over-indexed on turnout & voting access (7.9 points) and immigration (2.5).
Tone follows the same line. Non-US posts ran +14.0 net sentiment against +7.3 inside the country. The further from the ballot, the cooler the temperature — distance buys detachment, and the angriest themes are the ones only Americans were having.
Composition, not volume: each theme is a share of its own region's conversation, because the American side is 3.9 times larger and would otherwise swallow the other shape. Each spoke is labelled US share first, then outside-US.
Each bar is scaled to its own column's maximum, so the two rows read as rankings rather than against each other.
Each bar is the country's full post volume. The coloured segments are the themes above, and the two grey segments are the honest remainder: posts matching one of the other eight themes, and posts matching no measured theme at all. That last block is the largest almost everywhere, which is the finding — most of the conversation abroad was general election talk rather than any specific issue.
Themes overlap, so their raw counts are not a partition and cannot simply be stacked. The union of the themes is measured separately for each country, and the eight named segments are allocated inside that union in proportion to their own counts; hover any segment for its real post count. 177,872 further posts fall outside the United States but carry no country signal, so they sit in the non-US totals and not in this ladder.
Channel and geography are not independent. Twitter/X was 78% American, Bluesky 78%, Reddit 1% — but news coverage was only 58% American, because the world's press covered an American election. Read the night on X and you read a domestic argument; read it in news copy and you read an international story.
Engagement per post splits the same way: Facebook ran 313.1 and Twitter/X 0.00. Volume lives on X; reach lives on Facebook and Instagram.
Read these two charts against each other: the left one is attention, the right one is reach, and they do not agree. Donald Trump appears in 358,150 posts, 23.0% of everything collected, at -19.9 net sentiment — the most-named person of the night, and the most negatively discussed of those with enough labeled mentions to score. He was not on a ballot anywhere.
Behind him the order changes depending on what you count. Abdul El-Sayed is second by volume with 42,829 posts, but Abdul El-Sayed leads the field a reader would actually have seen. The takeaway for a story: being talked about and being read are different things, and a candidate who dominates mention counts can still be invisible in anyone's feed.
Treat the sentiment column carefully. 8 of the 20 names carry a thin marker: they were named often but drew too few sentiment-labeled mentions to score, so no number is shown rather than a number built on a handful of posts.
Counted by matching the person's name in post text with variants merged into one query, so "Trump", "Donald Trump" and "President Trump" are one row rather than three. Monitor's own entity panel reports different figures for these names because it is a separate annotation layer, not a post count; this page uses one measurement throughout so every number reconciles with the headline.
Each account below was classified by reading its own posting history, not its display name. This cut covers Twitter and Reddit, the platforms whose timelines the profiler can reach; every name links to the account. Ranked by engagement earned in the window.
95 of the accounts read were individuals: commentators, creators and activists rather than institutions.
22 were organizations: news outlets, campaigns, advocacy groups and one prediction market.
425 accounts could not be classified, most because they post more than the profiler samples; several are among the loudest in the window, so these leaderboards are not a complete ranking of the night: @nytimes, @hollywoodunlocked, @usatodaypolitics, @kamalaharris, @vanityfair, @FoxNews. A further 1 automated accounts were excluded by name (PoliticsModeratorBot), because a subreddit's automoderator is not a voice in the conversation.
Ranked by total engagements inside the window, each shown in a fixed-height tile so the grid stays even however long the post runs; scroll inside a tile to read the rest. Every author name and every View link opens the original.
nytimes2026-11-04 05:02
hollywoodunlocked2026-11-03 15:21
kamalaharris2026-11-03 15:15
nytimes2026-11-04 06:50
nytimes2026-11-04 02:15
cnn2026-11-03 14:33Engagement ranking finds the loudest posts, not a representative sample. No automated accounts appeared in the ranking — a subreddit automoderator's stickied thread collects engagement without being anyone's opinion. Links were live when this page was generated.
nytimes2026-11-04 05:02
kamalaharris2026-11-03 15:15
hollywoodunlocked2026-11-03 15:21
theshaderoom2026-11-03 21:5512:00 AM EST, 3 November to 12:00 AM PST, 4 November 2026 — 03 November 05:00 to 04 November 08:00 UTC, 27 hours. The window is absolute rather than rolling, so every figure describes one snapshot and the parts reconcile with the whole: US 1,163,192 plus outside-US 394,760 equals the total 1,557,952 exactly.
The boolean query of Quid Monitor topic 2957239 — a two-tier rule over all languages, scored at 0.863 precision against a thousand judged posts before this page was built. It is the query only: the saved topic also collects everything its 29 channel accounts post and a licensed news feed, which together add about half as much volume again, most of it not about the midterms. Twitter/X arrives as a 10% sample projected to full volume, so its counts are estimates and cannot be reconciled against individually retrievable posts.
Net sentiment is (positive − negative) ÷ (positive + negative) × 100; neutral posts are excluded from the denominator. Only 7.1% of posts carry a sentiment label at all (110,471 of 1,557,952), so every bar is a split of that labeled slice and not of the whole conversation. Any row under 250 labeled mentions shows thin instead of a score.
"United States" means a post Monitor located in the US. "Outside the US" is everything else, which includes 177,872 posts carrying no location signal at all; 216,888 carry an identifiable country and those are what the ladder ranks. Treat the non-US total as an upper bound and the country ladder as the floor.
Themes are boolean rules over post text, not a classifier, and they overlap. People are counted by name match with variants merged into one query. Authors are classified by reading their own posting history; accounts whose names mark them as automated are excluded from the post ranking. Age is absent: Monitor carries no age attribute, and inferring it from the loudest accounts returned no age for 92% of those readable, so no age figure appears anywhere on this page.
This is not a poll and carries no electoral inference. It measures what was posted in public, which over-represents people who post. It does not measure who voted, how they voted, or what the wider public believed. Coordinated activity is not filtered out.
Generated 04 November 2026 09:07 UTC from Quid Monitor. Metrics may not match the Monitor interface exactly. Read as exploratory.