— public posts from — accounts on Threads and X. — are in Malay or Malay–English, and Malaysians post most at — Malaysia time.
Words and two-word phrases whose share of posts this week is highest compared with all earlier posts, used by at least 15 different accounts. Indonesian posts are left out. Hover for the numbers.
The 25 busiest search topics over the last 30 days, posts per day by post date. Darker is busier; read each row against itself.
Search topics by average likes per post (topics with at least 20 posts). Likes are as counted when we collected each post.
Posts dated in the last 7 days, per search topic, both platforms.
Every search topic with at least 20 posts. Malay share is the part written in Malay or Malay–English. Click a column to sort.
| Platform | Topic | Posts | Last 7 days | Avg likes | Total likes | Malay share |
|---|
Share of each platform's posts by language, from a word-list detector tuned for Malaysian Malay. Very short posts can't be told apart.
Ranked by how many different accounts used each tag, so broadcasters' own brand tags don't crowd the list.
Posts by weekday and hour, Malaysia time (UTC+8), both platforms. Darker is busier.
By post date. Collection started on 29 October 2026, so earlier days hold only what search results and account histories reached back to; the jump at the end is our collection starting, not a surge in posting.
Topics ranked by replies and quotes per like (median over posts with 10+ replies and 5+ likes). A high ratio usually means disagreement or a pile-on rather than applause.
Search-result posts only, so a few huge accounts don't decide it. Average likes for each kind of post against that platform's average post: right of the line gets more, left gets less. Associations, not causes.
Approximate tokens (characters ÷ 4) by where posts came from and their language. Shows where the pretraining corpus is heavy and where it is thin.
Search topics with the fewest approximate tokens so far: candidates for more searching or related terms.
Groups of accounts that mostly reply to each other (accounts with 3+ reply links, grouped by who they talk with). The central accounts have the most replies inside the group; the topics are where its members turn up in our searches.
| Group | Platform | Accounts | Replies inside | Shows up in |
|---|
Followed accounts, typed from their name and bio by a small AI model. The number after each bar is how many accounts.
Time from a post to each reply we collected, as a share of each platform's replies.
Topics with the highest share of Malay–English mixed posts (topics with 50+ posts in Malay, mixed or English).
Posts using two or more marker words of a dialect (for example hang/depa/habaq for northern, kamek/kitak for Sarawak). A rough signal: short lists miss most dialect posts.
Malaysian slang words by how many posts use them. Hover for the number of accounts.
Read by a small AI model (Claude Haiku) on a sample of — posts: up to 30 of the newest posts per search topic and platform, in Malay, mixed or English. Expect mistakes on sarcasm and heavy slang; read these as directions, not exact shares.
Net mood: share of positive posts minus share of negative posts (topics with 15+ posts read).
Complaints by sector, as a share of all posts read.
Each complaint summarized by the model in a few words.
| Complaint | Sector | Topic | Platform | Likes |
|---|
Companies and products named in the posts read, with the net mood of those posts.
| Brand | Mentions | Positive | Negative | Net |
|---|
Parties and politicians named, and the stance of those posts toward them as the model read it. A small sample from our topics: not a poll.
| Name | Mentions | Support | Oppose | Neutral |
|---|
The most common themes the model gave posts within each search topic (count in brackets).
| Topic | Posts read | Main themes |
|---|
This is a sample of the Malaysian conversation, not all of it. Posts come from 212 Malaysian search topics (newest first on revisits), the accounts we follow (found by searching and by checking who writes in Malay), and the full history of larger X accounts. Topics we didn't search and accounts we don't follow are missing. Likes and replies are snapshots taken when each post was collected, so older posts have had longer to gather them. Retweets are left out; each post is counted once even if found several ways.