Updated 2026-09-18 · Chad

Sushiro Apologized Same-Day, a Pop Star Responded Next-Day: A Fast Official Reply Doesn't Mean the Story Has Cooled — Track the Timeline with LIHKG-scraper

In one line: Sushiro apologized in the comments the same day its conveyor-belt video went viral, and Chen Xiaodong responded publicly the day after his baseball-stadium appearance was cancelled — both official reactions were fast, but a news article only captures the moment someone responded, not whether the online discussion is still climbing or has already gone quiet. To see that timeline, schedule LIHKG-scraper to re-run the same keywords and compare how reply_count and last_reply_at change between runs.

This post contains Apify affiliate links (no extra cost to you). Costs and comparisons are stated honestly, including where Apify is not the best fit.

A fast official reply doesn’t mean the conversation stopped with it

On 2026-08-27, a video of a Sushiro employee crawling inside the restaurant’s tray-return conveyor belt went viral on social media. Sushiro’s official account replied the same day in the comments: “We have grasped the situation and are urgently launching an internal investigation to clarify the facts. We sincerely apologize for the concern and unease this has caused” (Liberty Times Net). The story didn’t end with the apology — Taipei’s health department inspected the branch that same day and found uncovered food in the refrigerators and an unclean floor, ordering fixes by August 30 or facing fines of NT$60,000 to NT$200 million under Food Safety Act Articles 8 and 44 (Liberty Times Net). From the viral video, to the apology, to the regulator’s deadline, that’s at least three separate moments spread across several days — not a single event that ended when the apology was posted.

The same week, a different story was circulating on Cantonese-language forums: on the evening of 2026-08-29, the TSG Hawks baseball team announced it was cancelling singer Chen Xiaodong’s scheduled opening-pitch and post-game performance slot at Taipei Dome, after fans pushed back over social media posts he’d made in 2019 during Hong Kong’s anti-extradition protests, including “love the country, love Hong Kong, oppose violence” (LTN Entertainment). The next day (August 30), Chen responded publicly, mentioning he’d be back at Taipei Dome on October 3 for a different event (TTV News).

What both cases share: the official or public-facing response came fast — same-day to next-day — but that’s only the single moment a news outlet happened to capture. Sushiro’s story kept running for days because of the regulator’s compliance deadline; Chen Xiaodong’s got another turn when the October 3 appearance surfaced. Reading one article, or scrolling a forum once, doesn’t tell you whether a discussion is still climbing or whether nobody’s replying anymore — that’s exactly the gap that scheduled, repeated measurement fills, instead of a single glance.

What a one-time check tells you vs. what a scheduled LIHKG-scraper run tells you

	Reading one article / scrolling once	Scheduled LIHKG-scraper monitoring
See the moment the story broke	✅	✅
See that an official response happened	✅ (if it made the news)	✅
See whether the discussion is still growing or has stalled	❌ just a single snapshot	✅ compare `reply_count` / `last_reply_at` across runs
Labor required	Re-check manually every time	Runs on a schedule; you just compare results
Cost	Free but time-consuming	Billed per row actually returned (pay-per-event)

How to track a topic’s discussion timeline with LIHKG-scraper

Step 1 — Use search mode with keywords, and set the order and time window for monitoring

{
  "mode": "search",
  "keywords": ["Sushiro", "壽司郎"],
  "order": "desc_reply_time",
  "dateFrom": "3d",
  "maxItems": 30
}

order: desc_reply_time (latest-reply-first) is the sort the input schema itself flags as built for monitoring — paired with dateFrom, it lets the run stop paginating early, which keeps a scheduled job cheap. dateFrom accepts relative time (3d = within 3 days, 12h = within 12 hours) or an absolute date; leave it blank to skip the filter.

Step 2 — Re-run the same keyword set on a schedule

Use Apify’s built-in Schedule feature, or an external scheduler like n8n or cron, to re-run the same input every few hours or once a day. Each run returns the current list of matching threads, with fields including thread_id, title, reply_count, created_at, last_reply_at, thread_url, and scraped_at.

Step 3 — Compare the same thread_id across two runs

If a thread’s reply_count is clearly higher than last run, and last_reply_at keeps getting closer to now → the discussion is still active, worth continued attention.
If reply_count hasn’t moved across several scheduled runs and last_reply_at is stuck far in the past → the discussion has gone quiet, and you can dial back monitoring frequency.

What this method can’t do: it only gives you the objective count of whether a thread is still getting replies right now — it’s not a prediction model, and it can’t see the motive or coordination behind any account. Judging whether a topic is likely to reignite, or whether the replies themselves are genuine rather than coordinated, still needs a human to review the actual content.

What it actually costs

LIHKG-scraper’s full set of billing events and prices is laid out in the early-warning piece’s pricing table; the search mode used here runs about $0.04-$0.06 per scheduled pull of 20-30 rows, or roughly $5-$7 a month at 4 runs a day.

Three ways to start

From least effort to most freedom

I run it for you Managed, daily to your sheet You do nothing. From US$299/mo. Use it here Run the free tool Instant, no setup, on this site. Run my actor On Apify, your account Same data, you run it and pay per result. Build your own Sign up for Apify Free account — build the pipeline yourself.

This post contains Apify affiliate links (no extra cost to you). Costs and comparisons are stated honestly, including where Apify is not the best fit.

FAQ

Did Sushiro really apologize the same day the video went viral? Yes. After a video of an employee crawling inside the tray-return conveyor belt went viral on 2026-08-27, Sushiro’s official account replied the same day in the comments: it had grasped the situation, was launching an urgent internal investigation, and apologized for the concern caused. Taipei’s health department inspected the branch that same day, found uncovered food in the fridges and an unclean floor, and ordered fixes by August 30 — non-compliance risked fines of NT$60,000 to NT$200 million under Food Safety Act Articles 8 and 44.

Did Chen Xiaodong really respond publicly the next day? Yes. On the evening of 2026-08-29, the TSG Hawks announced they were cancelling singer Chen Xiaodong’s scheduled opening pitch and post-game performance at Taipei Dome, after backlash over his 2019 social media posts made during Hong Kong’s anti-extradition protests. The next day (August 30), Chen responded publicly on social media, noting he would appear at Taipei Dome again on October 3 for a different event.

How should I configure LIHKG-scraper’s search mode for scheduled monitoring? Set mode to search, put the brand or event keywords you’re tracking into keywords, choose desc_reply_time for order (latest-reply-first), and set dateFrom (e.g. 3d) to limit results to threads that are still active. That lets a scheduled run stop paginating early and keeps costs down. Log thread_id, reply_count, and last_reply_at from every run and compare them against the previous run to see whether a discussion is still growing or has stalled.

Can this method predict whether a topic will flare up again? No. It only surfaces objective numbers — reply_count and last_reply_at — showing whether a thread is still getting replies right now. It isn’t a prediction model, and it can’t see the motives or coordination behind any account. Judging whether a topic is likely to reignite still requires a human to review the actual content.

Is scraping LIHKG legal? We only scrape LIHKG’s public pages, never login-walled data, and we follow the platform’s terms of service and local law.