The Great Recall: The Largest Recall Wave in Taiwan’s History, Measuring Every Tick of Democracy

30-Second Overview: In 2025, Taiwan did something rare in the world: it used the constitutional power of recall to hold a referendum-style accountability vote against more than a quarter of its legislators. In three waves of voting from July to August, none of the 33 recall cases passed—the largest recall wave in history removed not a single seat. But if you zoom out: this movement was more like a stress test of direct democracy. On the night of July 26, a 55.83% turnout set a new record in recall history, simultaneously measuring how high the recall threshold is and how much a polarized society must pay. It removed zero, but measured a great deal.

On the night of July 26, 2025, vote count results poured in from across the island. On this day, the fates of 24 Kuomintang (KMT) legislators and Hsinchu Mayor Gao Hung-an were decided at once by voters casting ballots—25 cases voted on simultaneously, the most recall votes ever held in Taiwan at one time. In the early evening, some districts saw consent votes take the lead, and voices were heard shouting in the recall groups' live streams. But by late night, as the 25 report cards were completed one by one, the conclusion was consistent: all failed¹.

The turnout that night reached 55.83%, a new high in Taiwan’s recall history². A civic mobilization of the largest scale in history yielded a result of "not a single seat removed." A night that removed nothing measured things that filled the entire negative.

This movement was later commonly referred to as the "Great Recall." To understand what it measured, we must first understand the ticks on the ruler.

Three Gates to Recall: 1%, 10%, and One-Quarter

In Taiwan, recalling an elected official requires crossing three numerical thresholds in sequence. The first is "proposal": the number of proposers must reach 1% of the total number of voters in the original electoral district. The second is "signature": the number of signatories must reach 10%. Only after crossing these two gates does the case become established and enter voting. The third gate is the vote itself—the consent votes must exceed the dissent votes, and the consent votes must also reach one-quarter of the total number of voters in the original electoral district for the recall to pass³.

These three gates also determine who has the opportunity to be put on the recall stand. Legislators elected via the National Not-at-District (NAD) proportional representation system are allocated seats based on party vote shares and have no "original electoral district"; institutionally, they cannot be recalled⁴. This is why all 31 legislators put up for recall votes in 2025 were district legislators. Those eligible to vote were voters aged 20 and above who had registered their residence in the district for at least four months. Eighteen-year-old high school students could not vote, nor could overseas citizens without Taiwanese household registration⁵.

The height of the third gate was adjusted to its current form only in 2016. Before that, recall adopted a "double 21" threshold: a turnout of over 50% and consent votes exceeding half of the valid votes were required for passage. In the 2015 "Cut the Tail" action recalling legislator Tsai Cheng-yuan, the turnout was only 24.98%, failing even the turnout gate, causing the recall to automatically fail⁶. In November 2016, the Legislative Yuan changed the threshold to "consent votes must exceed dissent votes and reach one-quarter of the voters," removing the hard requirement for a 50% turnout⁷. The proposal for this amendment came from several versions by the Times Party caucus and Democratic Progressive Party (DPP) legislators, passed after joint review.

Ironically, one of the parties that proposed lowering the threshold, the Times Party, saw its representative Huang Kuo-chang become the first legislator put on the recall stand after the new system took effect in 2017. That time, he received 48,693 consent votes, failing to cross the 63,888-vote threshold, and kept his seat⁸. Those who lowered the threshold became the first test subjects of the new system. That time, the threshold was not crossed. This debate on "how easily incumbents should be challenged" had already gone through a round before the Great Recall, laying the groundwork for the 25% threshold in 2025 (see also Taiwan Elections and Party Politics).

Source: Public Officials Election and Recall Act, Law Source Legal Network

The ruler is set. The next question is: why did the energy accumulate to the point of using this ruler? That starts with a difference of one seat in the Legislative Yuan.

A Legislature with a One-Seat Difference

The January 2024 legislative election produced a Legislative Yuan where no single party held a majority. The KMT won 52 seats, the DPP 51, the Taiwan People's Party (TPP) 8, plus 2 non-aligned seats leaning pan-blue, totaling 113 seats⁹. The KMT had only one seat more than the DPP, but through cooperation with the TPP, it placed Han Kuo-yu in the chair of the Legislative Yuan president on February 1 with a vote of 54 to 51¹⁰. A legislature with three parties under 50% has been a tug-of-war from day one.

The tug-of-war quickly spread from personnel to legislation. In May 2024, the KMT and TPP pushed through amendments to the Legislative Yuan Act, expanding the legislature's investigative and inquiry powers and adding a "Contempt of Parliament" crime. On the day of the third reading at the end of May, large crowds gathered outside the Legislative Yuan to protest, marking the starting point of the subsequent "Bluebird Movement." The Executive Yuan requested a review, which the Legislative Yuan rejected on June 21 by a vote of 62 to 51¹¹. The dispute went all the way to the Constitutional Court. On October 25 of the same year, Constitutional Judgment No. 9 ruled that several articles (the Contempt of Parliament crime, some inquiry and investigation provisions) were unconstitutional and invalid, but simultaneously ruled that while the legislative procedure had flaws, it had not reached the level of unconstitutionality. This was a "partially unconstitutional" ruling, not a total overturning¹².

The Legislative Yuan chamber during the power amendment battle on May 21, 2024. Photo: QuYi96 / Wikimedia Commons, CC BY-SA 4.0.

What truly fanned the flames was the night of December 20, 2024. The KMT and TPP jointly passed three laws in third reading: the Recall Act required recall signatures to include photocopies of ID cards front and back, and increased the penalty for forged signatures to a maximum of five years. The Constitutional Court Procedure Act raised the threshold for the Constitutional Court's deliberation to ten or more justices. The Revenue Distribution Act significantly adjusted the revenue distribution between the central and local governments; according to Ministry of Finance estimates, the central government would release an additional 375.3 billion NTD to local governments¹³¹⁴. One month later, on January 21, 2025, the Legislative Yuan also passed the third reading of the 2025 Central Government General Budget, cutting 207.5 billion NTD and freezing 260 billion NTD (these are two different actions and cannot be added together)¹⁵¹⁶. Over the entire session, the battles over these bills and budgets turned the legislature into a pressure cooker.

Before the ruler started reading, the ticks were already shaking. The pressure accumulated in the chamber for a year overflowed the walls, first turning into bluebirds outside the legislature, then into signature books mailed into mailboxes one by one.

Source: Central Election Commission, Legislative Yuan, The Reporter compilation

1,300 Signature Books in the Mailbox

The energy of the Bluebird Movement did not directly translate into recalls. Chen Hsiao-wei, the lead petitioner of the Taoyuan recall against Niu Hsi-tsing, is a pastor. He remembers that curve: "Taoyuan started collecting recall signatures last July, but within 4.5 months, the mailbox only received 9 signatures for Tu Chuan-chi and 11 for Niu Hsi-tsing. After the December winter Bluebird Movement, we could receive 100 to 400 per day. In early February this year, after the Lunar New Year, opening the mailbox revealed it was overflowing with over 1,300 copies."¹⁷ The sea of mobile phone lights on the streets took nearly seven months to settle into signature books with attached ID photocopies mailed into mailboxes.

On January 22, 2025, Foxconn founder Terry Gao (Tsao Hsing-cheng) mobilized the Southern Society, the Taiwan Professors Association, and other groups to establish the "Anti-Communist Taiwan Defense Volunteer Alliance," creating an umbrella organization for scattered local recall groups¹⁸. Terry Gao himself played two roles in this: convener of the alliance, and lead petitioner of the "Remove Black Core" team for the Hsiao Bi-khim recall case. The latter was just one of the more than 30 recall groups, not all of them.

Terry Gao speaking at a protest outside the Legislative Yuan on May 21, 2024; eight months later, he mobilized local recall groups to form the Volunteer Alliance. Photo: TMYAO / Wikimedia Commons, CC BY-SA 4.0.

The volunteers catching this wave looked different from before. Field observations by The Reporter found that the proportion of female volunteers in each district reached 70% to 90%, whereas in past recall efforts, volunteers were "at most 60-70%"¹⁷. The age demographic also shifted from the high school students of the 2015 "Cut the Tail" action and the fresh graduates of the 2020 anti-Kao campaign, to the main force of 2025: the generation in their 30s and 40s, described by interviewees as "almost all mid-level managers in various fields"—from the Sunflower Movement to the Bluebird Movement and then to the recall groups, a clear curve of street generations de-studentizing and moving toward middle age is visible (see also Social Movements and Civic Participation). Some recall groups had very strict vetting; "Mountain Remove Wei Harm" required applicants to record a video of themselves saying "Xi Jinping steps down" to be archived for review¹⁷.

This batch of new volunteers brought two faces. One is an efficiency-oriented organizational culture. Alliance advisor Chien Chia-yu described: "We used to discuss theory, talking with everyone 'why we should recall this legislator'; but now recall groups seem like companies, focusing on KPI thinking." The other is a very personal belief. A volunteer for the Banqiao "Great Delete Yuan" group, Bang Mang Mang, remembered that when she said she was going to volunteer, her husband initially asked: "It won't pass, right? What can you change alone?" Hsiao Shih, a Hong Kong immigrant and co-petitioner for the Taoyuan recall group, said that participating in the recall was the third time she felt like a Taiwanese. The first sentence of Taiwanese she learned was "Recall saves Taiwan"¹⁷.

The cultural sector was also swept up. Writer Yang Shuang-tzu launched a signature campaign that grew from over 200 people in early February to 1,043 writers, with the signature statement quoting Brodsky: "Literature has the right to interfere in politics, until politics stops interfering in literature"¹⁹.

The first tick of the ruler came out: the mobilization energy was real, and it set a record. With great energy, both sides started using the system as a weapon against each other.

Who Launched the Great Recall: Different Answers from Both Sides

The most basic thing about this movement is who launched it, and the two camps gave completely different answers.

The self-narrative of the recall groups is: this was a citizen-initiated action independent of the DPP, with the core axis being "anti-communist, defend Taiwan." This statement was also echoed by some external observers; the University of Nottingham's Taiwan Insight academic blog described it as a "citizen-initiated" movement, not just a party confrontation²⁰.

The opposing narrative is different. KMT legislator Hsiao Bi-khim pointed the finger at the ruling party in her recall defense book, alleging that "DPP Legislative Yuan party leader Ko Chien-ming repeatedly admitted in person that the Great Recall action was planned and launched by him hand-in-hand," and called it a "fake civic movement manipulated by the DPP"²¹.

It must be noted that this is Ko Chien-ming's narrative-style allegation against Ko Chien-ming in Hsiao Bi-khim's defense book, belonging to the opposing camp's narrative. As for Ko Chien-ming's own publicly spoken, verifiable related remarks, there is only one sentence after the vote on August 24: "The Great Recall is still a great success; 1.3 million signature books prove that public opinion has risen"²². The recalled legislators mostly responded with performance records; Fu Kun-chi listed in his defense book "77.2% governance satisfaction" and "created the only five-star nine-peat in the country"²³. KMT Chairman Chu Li-lun set the tone at a rally in Banqiao on July 19: "No ruler in Taiwan's history has ever launched a great recall against the opposition; you cannot find any democratic country in the world... Only President Lai dares to do this"²⁴.

What can be confirmed is only the structural fact: for the 33 recall cases, the statutory proposers and lead petitioners were all individual citizens. The KMT's "recall to check recall" action against DPP legislators came from a resolution of the KMT party caucus meeting²⁵. The forms of launch were different; as for whether this counts as "party manipulation," the judgment is left to the reader.

The People's Party stood in its own position. They publicly opposed this action they called "evil recall," but also refused to cooperate with the KMT to raise the recall threshold²⁶. During the voting period, party chairman Ko Wen-je, who was detained in the Tucheng Detention House, received a recall voting notice. Huang Kuo-chang asked on Facebook: "May I ask Lai Ching-te: what kind of black humor is this? Notifying Ko Wen-je that he has the right to vote, but I won't let him vote!"²⁷

The second tick of the ruler starts reading from here—measuring the depth of polarization: halfway through the test, there was no consensus on who pressed the start button. And the most fiercely fought battlefield was the signature books themselves.

1,784 Dead Signatures vs. 160 Forgeries

Let's clarify one thing first: the signature books of both camps were found to have problems, but the two ledgers record different things.

One is "dead signatures": people who have already passed away appear in the signature book, the problem lies in the清查 (check and comparison) of household registration data, not necessarily involving active forgery by anyone. The other is "forgery": someone used another person's personal data to forge signatures to fill out signature books, which is an active act of forgery. The legal nature of the two is different and cannot be conflated.

In April 2025, the Central Election Commission (CEC) sent 41 recall cases involving forgery or dead signatures to the Supreme Procuratorate for prosecution, of which 15 were cases recalling pan-blue legislators, 17 recalling DPP legislators, 8 recalling pan-green councilors, and 1 was the Gao Hung-an case²⁸. Here is a口径 (口径/trap) that must be laid out: the same thing, calculated at different stages of verification and using different units, rolls out several numbers. By "copies," the 17 cases recalling greens involved 1,784 dead signatures and 42 forgeries in the first-stage signature check in April. The cases recalling blues had 12 dead signatures and 160 forgeries, with Lin Pei-hsiang's case having the most at 116 forgeries²⁹. But by June 21, when the CEC officially certified, the unit changed to "cases"—16 cases involved dead signatures, and 12 involved forgeries³⁰. "Copies" and "cases" are not the same thing; any number must be labeled with which stage and which unit.

The denominator must also be looked at together: all 17 cases recalling greens were found to have dead signatures, while among the 35 cases recalling blues, only 15 were involved. These numbers placed together allow readers to judge their meaning themselves; this article does not draw a conclusion for them.

One case has reached a conviction. Chuang Chan-kuei, deputy chairman of the KMT Tainan City Party Committee, along with 11 other party workers, used party member lists to fill in personal data without consent and forge signatures to make up 3,000 proposal copies each for the recalls of Lin Chün-hsien and Wang Ting-yu within ten days, totaling 3,663 forgeries. On July 10, 2026, the Tainan District Court sentenced Chuang Chan-kuei to two years in prison with a five-year probation, and required payment of 600,000 NTD to the public treasury under the Personal Data Protection Act. The other ten received sentences ranging from six months to one year and seven months, with two to five years probation. The judgment reasoning pointed out that such behavior not only infringes on the information privacy of the forged but also destroys the legitimacy of the democratic recall system³¹. It must be distinguished that what was prosecuted was the act of forgery; ordinary citizens participating normally in signatures do not involve criminal liability.

The KMT questioned the prosecutors for "handling blues but not greens," pointing out that the second-stage signatures for the recalls of blue legislators totaled over 150,000不合格 (unqualified), asking why they were not investigated with the same force³². The CEC responded: "As long as there are pre-signature deaths or forgery incidents, we will依法 (according to law) follow routine procedures to report for prosecution," emphasizing "no differential treatment"³⁰. At the time of this article's verification, convicted and prosecuted cases were concentrated on the KMT party worker side. But this may also reflect that most green-recall cases were not sent for prosecution, resulting in a smaller sample base; both interpretations require more complete judicial statistics to conclude.

Source: Yahoo News March 2025 report (1,543), ETtoday March 14, 2025 (1,737), Liberty Times April 17, 2025 (1,784)

The second tick is completed here—the cost of polarization: even the numbers have two ledgers, and social trust depreciates along with it. The battle raged for half a year, and finally, supporters, opponents, and the accused all walked into the same polling station.

PTS Evening News report on July 23, 2025: Three days before the July 26 vote, pro and con camps took turns fighting for volume at roadstops, urging dissent and consent votes.

Consent Votes Crossed the Threshold, but Still Lost to Dissent Votes

The voting was divided into three waves. The first wave was July 13, the recall case of Nantou County Councilor Chen Yu-ling—this was the only case that became established under the KMT's "recall to check recall" strategy, targeting the green camp, opposite in direction to the next two waves. The result was 12,160 consent votes and 5,867 dissent votes; the consent votes did not reach the one-quarter threshold of voters, and it failed³³.

The second wave was the 25 cases on July 26: 24 KMT legislators plus Hsinchu Mayor Gao Hung-an. The national turnout was 55.83%, a new high in recall history. The consent mobilization rate was 23.75%, dissent 32.08%². All 25 cases failed.

The motives for those 2.18 million dissent votes were not singular: there was the base supporting their own district's legislator—these districts were places the KMT had won in 2024; there was reservation toward recall as a political tool; and there was the exhaustion of wanting life to return to normal.

In this wave, the consent votes of seven cases actually crossed the 25% threshold, but the dissent votes were higher, keeping the seats. The vote margins for these seven people formed a spectrum: the most precarious defense was Yeh Yuan-chih in New Taipei, with only 3,560 more votes. The widest gap was Cheng Cheng-chien in Hsinchu, with 29,335 more votes³⁴. Even though they all "crossed the threshold but were not recalled," the tightness of their fates differed by more than eight times.

Source: Central Election Commission, CNA

Source: CNA

Gao Hung-an's case had two clocks running. She was sentenced guilty in the first instance for corruption and suspended from office under the Local Government Act; she was a "mayor on suspension" during the recall vote. This recall consent was 86,291 votes, dissent 124,360 votes, and it did not pass³⁵. If the recall had passed, she would have been removed from office directly and would not have been reinstated even if the case was overturned later. The fact that the recall failed instead preserved the space for her to wait for the judicial result—on December 16, 2025, the second instance changed the corruption part to a not-guilty verdict, allowing her to apply for reinstatement³⁶. The political clock and the judicial clock intersected on one person for a year and a half.

The third wave was the seven cases on August 23, all KMT legislators, voting on the same day as the Nuclear 3 extension referendum. It must be clearly distinguished: on July 26, there were "7 cases that crossed the threshold but were still vetoed"; on August 23, it was another 7 people, 7 cases that "didn't even cross the threshold." The two sets of seven cases were completely different people. The Nuclear 3 referendum on the same day received over 4.34 million consent votes, accounting for 74.17%, far more than dissent votes, but failed to pass because it did not reach the national voter one-quarter threshold (about 5 million votes), with a turnout of only 29.53%³⁷.

The vote count process that night was live-streamed by TTV News in its entirety. The third tick of the ruler appeared on voting day: the threshold is not an abstract number; it is 2.18 million dissent votes. After voting, the bills and reports were just beginning to arrive.

The Battlefield Moves Back to the Legislative Yuan

The voting ended, and the stress test report was released. This report has four pages: the bill, the stalemate, the laws not yet amended, and the external reading.

The first page is the bill. The election expenses for the 31 legislative recall cases totaled approximately 400 million NTD; the Nuclear 3 referendum on the same day cost another 1.14 billion NTD; the Gao Hung-an case's 31.01 million NTD was funded by the Hsinchu city government's local budget³⁸. A movement that did not recall a single person sent out a stack of invoices first.

📝 Curator's Note: Reading the Great Recall as "a failed mobilization that changed nothing" is the most convenient通行 (common) narrative—the map didn't move, seats didn't change, but the bill was sent to the whole table. But this narrative treats "results" as the only ruler, exactly missing what the stress test was truly doing: it measured how much weight a bridge can bear. Zero war results and record-breaking turnout coexist; this is the most honest set of readings for this test.

The second page is the stalemate. Ko Chien-ming called the result "the Great Recall is still a great success"²². Critics within the green camp, You Ying-lung, chairman of the Taiwan Public Opinion Foundation, described it as a "disastrous democratic civil war," demanding the ruling party apologize to society³⁹. The political map actually did not change: Lai Ching-te announced "four adjustments"⁴⁰; the Zhuo Ying-tai cabinet reshuffled 16 cabinet members⁴¹; Cheng Li-wen was elected KMT chairman in October⁴²; and Ko Chien-ming, amidst review criticism, completed his term and handed over to Tsai Chi-chang in February 2026 as usual⁴³⁴⁴, but the opposition alliance still held the legislative majority. The most concrete consequence of the stalemate fell on livelihood: the 115th Central Government General Budget crossed the year for the first time without being paid, and government policy execution was stuck⁴⁵.

The third page is the laws not yet amended. The threshold articles that caused the greatest controversy, such as "recall consent votes must be higher than the number of votes for election," were not addressed in the late 2024 amendment wave⁴⁶. By July 2026, when this article was verified, the related drafts were still listed on the Legislative Yuan's pending list and had not entered the plenary agenda⁴⁷. Brookings scholar Kharis Templeman judged after the vote that the "Pandora's box" of the recall tool has been closed again, and it will be used less in the future²⁶. The box may have closed, but the lock of the threshold hasn't been changed.

The fourth page is the external reading. International analysis mostly focused on the "executive-legislative stalemate": the government was divided before and after the recall. Templeman warned, "Taiwan cannot afford another 30 months of political stalemate"²⁶. As for external reactions, they must be viewed independently and clearly marked as their nature—this is the stance of foreign governments on Taiwan's internal affairs, not listed alongside Taiwan's internal perspectives. The Taiwan Affairs Office of the PRC spokesperson repeatedly called the Great Recall "green terror" in the island starting from April 30⁴⁸. The Mainland Affairs Council (MAC) refuted this on July 23, pointing out that Beijing was "赤裸裸 (nakedly) intervening publicly, without disguise"⁴⁹. After turning the last page of the report, there are still things that numbers cannot measure.

Taiwan Developed on the Negative

Three waves of voting, 33 cases, none passed.

In terms of war results, this largest recall wave in Taiwan's history was almost empty. The number of legislators successfully recalled in Taiwan's constitutional history remains only one to this day—Chen Po-wei in 2021⁵⁰. Calculated by "how many people were recalled," the limit of this tool is clear.

But on the same negative, other things developed. The 55.83% turnout on the night of July 26 was a new high in recall history. Observers from the Taiwan Democracy Lab found that as recall groups mobilized, they gradually realized that the "anti-China, defend Taiwan" narrative was too far from the general public, so they shifted the focus of dialogue to livelihood issues like rental subsidies, long-term care, and education, actively contacting people of different layers. This was uncommon in past election battlefields⁵¹. This entire generation, which learned to organize and also learned to speak with people of different stances, where will they go after the movement exits? The Reporter still lists it as an open observation topic today; no one can write the answer dead for them⁵².

The authors of Taiwan Insight borrowed an old saying from 170 years ago to summarize that complex mood. In 1848, U.S. President Polk (James K. Polk) described the territorial expansion after the Mexican-American War: "Everyone was glad with it, but no one was proud of it." The author said this sentence holds true for Taiwan today as well⁵³.

On the late-night vote count screens, the three characters "Not Passed" lit up one by one. The next day, the signature tents were taken down one by one. What remained was the record-breaking turnout number and a whole generation of volunteers who learned for the first time how to organize strangers. Thirty-three "Not Passed" were the result of this test. The true value of the test lies in that it measured so much at once—the energy of mobilization, the height of the threshold, the cost of polarization. What was measured ultimately exceeded what was recalled.

Further Reading:

• Taiwan Elections and Party Politics — From the Zhongli Incident to the three-party standoff, the long-term evolution of Taiwan's electoral system and voting culture
• Democratization — Before the recall power was written into constitutional daily life, that generation's journey from martial law to direct elections
• Taiwan's Democratic Transition — How authoritarianism gradually handed ruling power back to the people
• Social Movements and Civic Participation — From the Sunflower Movement to the Bluebird Movement, the mobilization genealogy of street generations
• Taiwan's Political Environment and Electoral System — The institutional depth of the recall system and threshold design

Image Sources

This article uses 3 CC-licensed images, all cached in public/article-images/history/ to avoid hotlinking to the source server:

• Bluebird Movement Crowd Outside the Legislative Yuan 2024 — Photo: Fujifilmuser Taiwan, 2024-05-28, CC BY-SA 4.0, Wikimedia Commons
• Legislative Yuan Chamber 2024-05-21 — Photo: QuYi96, 2024-05-21, CC BY-SA 4.0, Wikimedia Commons
• Terry Gao 521 Rally Speech 2024 — Photo: TMYAO, 2024-05-21, CC BY-SA 4.0, Wikimedia Commons