Frozen Garlic

Voting for the 24 recalls was held Saturday, and they resulted in an unambiguous victory for the KMT. None of the 24 KMT legislators lost their seat. Only one of the recalls was even close. After a year in which the KMT bungled its response to the recall efforts again and again, it produced a shockingly triumphant result in the final and decisive juncture.

If you dig deeper into the outcomes, there are some more nuanced aspects to the results. I will discuss a few of those later in this post, but the biggest and most basic takeaway is that the KMT had a very good day and the recall campaigns failed utterly to achieve their goal of recalling any KMT legislators.

What caused this result? Pundits from all sides are busy spinning the outcome to tell you what it all means. In the aftermath of every election, you will hear a pundit passionately insist that, if they had only done what he proposed, the results would have been very different or that, precisely they because they did do what he favored, they got a successful outcome. And going forward, the only way they get a better result or continue to get good results is to pursue the strategy that he prefers.

I’m going to give you a far less satisfying answer. I don’t think there is one clear factor drove this outcome. Society is diverse, and different things mattered to different people. There are probably 20 different explanations, and, to some degree, they are all right. We don’t have any exit polls or other good objective evidence about what drove voting behavior; everyone is merely speculating. Some of the speculation is more reasonable than others, but no one has a definitive answer.

So let’s look at several of the proposed explanations, trying to remind ourselves that each one is probably somewhat right and somewhat incomplete.

• This was not surprising. These were all blue districts that elected KMT legislators, and they continued to be blue districts that supported their KMT legislators.
• This was not surprising. The KMT and TPP won 57% of the party list vote in 2024. They represent mainstream public opinion.
• Voters saw the recalls as a cynical DPP power play to overturn the 2024 election, and they rejected it.
• Voters did not want to give the DPP unified control of government. They would prefer the legislature to act as a check on DPP power and to effectively oversee the executive branch. They also worry about DPP dictatorial tendencies.
• Voters don’t want to vote every year. They voted in 2024, and they didn’t want encourage recalls and perpetual election campaigning. They just want things to return to the normal, non-election environment.
• This was an abuse of the recall provision. Recall should be done on a case by case basis rather than against an entire party, and perhaps only against legislators accused of corruption or other crimes.
• Voters are sick of hearing about Chinese infiltration. KMT sympathizers, in particular, are sick of being accused being willing or unwitting lackeys of the CCP.
• This was a vote of no confidence in President Lai.
• Lai and the DPP didn’t get fully involved, leaving too much of the heavy lifting to the recall groups rather than fully mobilizing all their party resources
• Voters were enticed by the prospect of a NT10,000 universal stipend
• Voters were repelled by the DPP’s rejection of the death penalty for child abusers.
• Voters are used to seeing majorities in the legislature ram their legislation through. The KMT/TPP hardball tactics to pass the legislative reform bill, the budget cuts, the Constitutional Court reform, and so on were not seen as fundamental violations democratic processes.
• The balance of between the executive, legislature, and Constitutional Court Is too abstract for ordinary voters to care about.
• Voters were not bothered by the budget cuts, and they were happy with reallocation of money from the central government to local governments.
• Ko Wen-je has been detained for nearly a year while prosecutors investigate a corruption case. This duration is seen as unfairly long and motivated by political pressure from the DPP. It has also cemented the alliance between the KMT and TPP, giving TPP voters a reason to come out to vote against the recall.
• The KMT legislators were strongly supported popular mayors, and this was a show of support Chiang, Hou, Chang, and Lu.
• I’m sure I’m overlooking half a dozen more arguments.

The narrative that’s starting to slowly gel in my head goes something like this. Late last year, the recall groups were organizing credible campaigns against several of the most controversial legislators. At that point, it was possible to argue that these particular legislators had violated fundamental democratic norms and needed to be replaced without necessarily arguing that the entire KMT/TPP coalition needed to be replaced. However, simply replacing 4 or 5 controversial KMT legislators with a different set of KMT legislators would not have solved the DPP’s problem of being a minority inside the legislature. For that, they needed to flip some seats, and that required attacking legislators in marginal seats rather than the most controversial legislators. However, once the DPP announced a universal recall campaign, the underlying logic shifted to a more fundamentally partisan clash. In a nutshell, it became very difficult to argue that KMT sympathizers should recall someone controversial like Wang Hung-wei and not worry about the partisan implications because they could replace her with another KMT legislator while simultaneously arguing that they were trying recall KMT legislators everywhere in order to change the balance of power in the legislature. The result of weakening this argument against controversial legislators was that the only viable path to success was a significant shift in public opinion away from the KMT and toward the DPP. In January and February, when the die was cast, a shift in public opinion didn’t seem too far-fetched. Polling showed that support for Lai and the DPP was holding up fairly well and the numbers for the KMT and TPP were lower than in January 2024. Critically, if this was turning into a partisan fight, the DPP needed to go all out. They needed to send out all their big guns, including the president, premier, mayors, all major party figures, and all legislators to all the marginal districts in an effort to change a few minds in those critical districts to transform them from light blue to light green districts. Instead, the big guns mostly stayed in the background, leaving the spotlight to the recall groups. In the end, the recall results did not show any evidence of a significant shift in public opinion. The results in July 2025 look a lot like the results in January 2024.

Of course, this is the narrative that I’m developing with the benefit of hindsight. None of this seemed obvious in the moment. Oh, and most of those other factors that I listed above also had an impact.

Whatever the reason, the indisputable fact is that the KMT turned out more people than the recall movements. One way to illustrate just how thorough the KMT victory was is to think about the 25% threshold. The No vote exceeded the 25% threshold in all 24 districts, while the yes vote only exceeded that threshold in seven cases. That is, even if no KMT supporters had turned out, only 7 legislators would have been recalled. The yes side simply did not turn out enough voters to achieve its goals. Meanwhile, the no side turned out more than enough. From this perspective, the outcome was both a KMT success and a recall failure.

One of the premises of recalls is that voters delegate political power to politicians in general elections to act on their behalf, and, if they are dissatisfied with the performance of those politicians, they can withdraw that delegation through a recall. This suggests that a recall should either show some evidence of voters changing their minds or of a decisive shift in passion and energy (or both). Some people who originally supported the politician should now regret that decision, or a large number of voters who could not be bothered to turn out in the previous election must be mobilized (or previous voters must become apathetic). We did not see any systematic evidence of this kind of shift in public opinion in these recall results. In general, voters reaffirmed the choices they made in 2024.

The first and second stages of the recall process involved petition drives. In these stages, people who were dissatisfied with the legislator were mobilized to sign petitions. The overwhelming majority of these people were never supporters of the legislator. These petitions represented high levels of anger and disgust among the part of the electorate that opposed the legislator in the first place. In the 3^rd stage, the supporters had their chance to weigh in. It turns out, they did not feel the same way as all those people who signed petitions.

(We didn’t see any evidence of a shift in public opinion in the other direction either. There isn’t much reason to believe that people who voted for president Lai or DPP legislators regret those choices.)

If one judges the recall campaign by whether they were able to recall legislators – and that was their stated objective – these recalls were a total failure. However, perhaps we should take a broader view of this process. There are some ways in which these recalls impacted political behavior and might affect strategic calculations going forward.

I doubt any of the KMT legislators particularly enjoyed going through this recall experience. They all survived, but they had to fight to keep their seats. They had to spend a lot of time and energy – and probably a bit of money – to mobilize their supporters for a vote that none of them asked for. I’m quite sure that they would have preferred to spend the last few months doing something else with their time and energy. It’s possible that the lesson they and future legislators take away is that they don’t have to worry about recalls, but it’s also possible that the enduring lesson will be to try to avoid the stress of recalls by behaving in less controversial ways. Having to fight a recall is, in and of itself, a significant penalty.

There is a bit of anecdotal evidence that KMT legislators changed their behavior during the last six months as the gravity of the recall threat sunk in. There has been far less controversial news coming out of the legislature in the last few months than in 2024. The majority coalition has focused its efforts on more popular legislation such as the NT10,000 stipend or the bill dealing with child abuse and less blatantly partisan legislation seeking to alter the fundamental balance of power between the branches of government. We have heard stories about legislators spending more time going home to do constituency service. One vivid example of this is Lee Yen-hsiu’s decision to resign her position in the KMT leadership in order to spend more time in her constituency fighting the recall. Inside the legislature, there have also been stories about toned down rhetoric. Legislators who a year ago were yelling and screaming at government ministers have been much more polite and restrained in recent months.

The recalls have effectively changed the calculations involving the political calendar. In 2024, the next general elections were nearly three years away. The majority coalition seemed to believe it could act with impunity since voters would probably forget about things in 2024 by the time late 2026 or early 2028 rolled around. The recalls turned 2025 into an election year, and legislators had to adjust their behavior. One might think that, since the shackles are now off, they will go back to their previous behavior. However, it will not be long until the 2026 election season kicks off in earnest, so legislators might have to be might have to consider the effect of their behavior on that election. In other words, the recall campaign may have effectively cut a year out of the “normal” times in which legislators don’t worry too much about public opinion.

What about those controversial legislators? Earlier, I suggested that the argument against them had been weakened by the transformation from a few individual-based recalls to a broad-based partisan recall. Even so, there is evidence that these controversial legislators underperformed compared to other KMT legislators.

In this table, I look at the ratio of the percentage of no votes in the recall to the vote of Hou and Ko in the 2024 presidential election. (Since the results for Taitung and Hualien are skewed by large numbers of Indigenous voters, I use my numbers for non-indigenous voters derived from ecological inference estimates for those two counties.) This table is sorted from lowest to highest, so the people at the top of the table did the worst relative to previous partisan patterns.

In a previous post, I (subjectively) sorted legislators into various buckets. One of the buckets was controversial legislators. The eight controversial legislators are identified with an asterisk*. Earlier, I mentioned that the yes vote against seven legislators passed the 25% threshold. These seven legislators are identified with a hashtag#.

All eight of the people I identified as controversial are found in the top half of this table, Fu Kun-chi, Hsu Chiao-hsin, and Wang Hung-wei, arguably three most controversial legislators, are in the top four. This suggests that were less successful at absorbing KMT and TPP support the Ian the less controversial KMT legislators facing recall. They all survived the recall, but, based on how everyone else did, it was probably closer than it should have been for these controversial legislators.

The Yes vote reached the 25% threshold against 5 of controversial legislators. For two of the other controversial legislators, Liao Wei-hsiang and Lo Chih-chiang, the yes vote was less than 1000 votes away from reaching the threshold. Most KMT legislators could have relied on low turnout to defeat the recall. If the controversial legislators had relied on low turnout, most of them would have lost their seat.

It might not be stark and vividly obvious to the casual observer, but the voters did not reward bad behavior in the legislature this time.

I want to make one final point about the recall process. In the runup to voting, I was asked several times whether the recall had polarized Taiwan’s society. I pushed back against this argument as strongly as I could. I do not believe the recalls fostered polarization and division.

In May 2024, events in the legislature were so extreme that many people felt the need to go out into the street to protest. Throughout the rest of the year, events in the legislature fueled these feelings of angst, anger, despair, fear, disgust, and betrayal. All that passion had to go somewhere. What did not happen is as important as what did. Taiwan did not experience widespread political violence, riots in the street, or police crackdowns to maintain order. The recalls funneled all that energy into legal process within the system. Volunteers spent countless hours learning the minutiae of the petition process, training themselves to talk to voters, sitting outside collecting signatures, checking petitions for errors, compiling them into the right format to submit to electoral commissions, and then working to convince voters to vote yes in the recall election. It all amounted to an enormous expenditure of time and energy. At the end of the process, people voted. The results did not come back the way the volunteers had hoped, but there was an objective result that the volunteers could not argue with. They didn’t fail because of some hidden conspiracy; they simply didn’t convince enough voters with their arguments. Democracy provided a constructive outlet to absorb and diffuse all that passion.

Is society completely unified today? Of course not. No society is. The point of democracy is that people have different values, goals, interests, and desires, but we can manage those differences through institutionalized mechanisms without resorting to violence. I think the recall campaigns were a successful example of this.

Here is a podcast and a webinar that I’ve done recently.

It’s a challenge to get all gussied up for video when you are disabled, and you might notice that I’m wearing a blue T-shirt in both events rather than something more formal. I’ll leave it to you to speculate whether my blue T-shirts are subtle gestures of support for the KMT, the Bluebird movement, or just that I like the color blue.

The Central Election Commission has announced the number of eligible voters in each district facing a recall as well as the 25% threshold needed for a recall to pass. Remember, the recall needs to satisfy two conditions. First, it must pass the 25% threshold for Yes votes. For example, The first row shows Taipei 3. In that district, at least 68,578 voters must vote to recall Wang Hung-wei. Second, there must be more yes votes than no votes. So even if there are 70,000 yes votes, Wang can defeat the recall by mobilizing 75,000 no votes.

So now when you watch the votes come in on Saturday evening, you will know what the thresholds are. Watch like a pro!

By the way, you can see the enormous disparity in the population of some of the districts. When the districts were drawn in 2007, all were within 15% of the city or county mean. They were supposed to be redistricted every ten years to keep the populations relatively equal. However, in 2017, the Central Election Commission and the legislature neglected to take up this responsibility and maintained the same districts wherever possible. So now, almost 20 years later, we do not have one person, one vote. New Taipei 1 has almost twice as many people as New Taipei 7, so people in New Taipei 1 only get about half a vote. It boggles my mind but those people aren’t screaming about this injustice. Hopefully, we will get a thorough redistricting in 2027 to rectify this egregious malapportionment.

I’ve been meaning to write about the polling results and the recalls, but I didn’t get around to it in time. Unfortunately, I had a major computer disaster and have spent the most of the last two weeks trying to fix my old computer and then buying and setting up an entirely new one. There were some interesting polling results about individual districts, but we are now in the polling blackout period so it is illegal for me to write about those. Sorry.

Five more recalls officially passed the second stage, so after the 24 recalls on July 26th there will be 7 more on August 23rd. However, not all recalls are equal. There is a reason that those 7 took longer to pass the second stage. Most of the recalls that are likely to pass will take place a week from today. (Remember when I was so confident several of the recalls wouldn’t pass second stage that I didn’t bother including them in my previous tables? Oops.)

A lot of the public focus during this campaign has been on a few districts with high profile legislators, especially Hualien (Fu Kun-chi), Taipei 3 (Wang Hung-wei), Taipei 6 (Lo Chih-chiang), and Taipei 7 (Hsu Chiao-hsin). Those districts are important since a successful recall or even a close call would send a strong message to all legislators about what is acceptable behavior in the legislature. However, if those districts are actually in play, we are probably looking at a historic result. We should probably pay more attention to the other districts.

I want to comment on a different race, New Taipei 1, that might have a different sort of impact. In 2024, Hung Meng-kai won this district by nearly 20%. It was a fantastic result for a district that historically has only been slightly favorable to the blue side. Of course, the population of this district is growing very fast — it is now the largest district in the entire country — so we don’t really know what the partisan balance is these days. Hung is now in his second term in the legislature and is very-well positioned to run for New Taipei mayor next year. We haven’t had a lot of chatter about the KMT nominee for New Taipei mayor, but I’m not impressed with the people that have been mooted so far. The most prominent is a deputy mayor Lee Szu-chuan, an old KNT War Horse Who has been appointed to several high positions but has never run for office. Apparently, they think they can repeat the model of Hou Yu-ih, who was deputy mayor before being elected mayor. However, he was a unique figure who had a national profile long before he became deputy mayor. Lee is basically a faceless party hack. The KMT would do much better to pick an elected representative. The KMT has six legislators in New Taipei, but two are old and radical and three are in their first terms. Hung is handsome, charismatic, not too inexperienced, and apparently fairly popular. To me, he is far and away the best choice. These advantages would be highlighted if he were to decisively defeat the recall campaign against him. On the other hand, if the recall were successful it would probably crush his hopes for the nomination.

A successful recall would wreck the careers of many of the incumbent legislators. For example , if any of the six legislators in Taoyuan were recalled, I wouldn’t expect to hear anything about them ever again. However, that is not true for everyone. A recall would be a significant bump in the road for someone like Hsu Chiao-hsin, but I wouldn’t expect it to end her career. She is young and charismatic enough that she could go back into the Taipei City Council and wait for another opportunity to move up. Similarly, Hung Meng-kai might still have a future, but this next year is a golden opportunity to become mayor and get on the ladder to the highest offices. That is to say, he is one of the legislators who has the most at stake in this recall campaign.

 (Regardless of who the KMT nominates, I think the person best positioned to win the New Taipei mayoral race next year is the DPP’s Su Chiao-hui.)

Next topic

In my previous post, I sorted the various districts into different buckets. One of the criteria was the partisan balance of each district. I thought I might do this in a more systematic way. The next two tables show each district’s result in the previous two presidential elections, sorted by the DPP’s vote share, going from most competitive to least competitive.

The two tables reveal similar rankings, but there are some interesting differences. The only district in green territory is Yunlin 1. New Taipei 7 is roughly similar to the national balance overall. These are districts that the DPP needs to win if it has any hope controlling a majority in the legislature. All the other districts are some shade of blue.

If you (arbitrarily) consider any district within 3% of the national average as competitive, there are 12 competitive districts on the first list and 14 on the second list. 9 are on both lists. That’s quite a few light blue districts.

One of the interesting results was that Taipei 3 and Taipei 4 were very similar in both elections but they were much more competitive in 2024 (ranking 4^th and 5th) than in 2020 (15^th and 16th). In the legislative races, Taipei 4 has been very close the last few cycles while the DPP has never put up a serious challenge in Taipei 3. I’m not entirely sure what to make of this. On the one hand, you might conclude that Wang Hung-wei is in serious trouble both because she is controversial and because her district is a lot closer than everybody thinks. On the other hand, you might think that they should both be safe unless they are so controversial that they have become toxic.

From years of observation, I think of Taoyuan 1 and 2 as very competitive or even green districts. However, the DPP did very badly in them in 2024.

The recalls in Taichung 3 and 8 are not until August 23, so I’ll save my commentary on those districts until later. For now, let’s just note how competitive those two districts look even though most observers think that those two legislators are among the safest of the KMT incumbents.

The figures for Taitung and Hualien are skewed a heavy population of Indigenous voters who don’t affect the district recalls. We don’t have official data on how the non-indigenous voters voted in the presidential elections, but my unofficial estimates are that Lai got about 34% in Taitung and 29% in Hualien. That makes the former solidly blue and the latter deep blue. These are not districts the KMT should be having trouble with.

Last week the central election Commission announced that 24 recalls will be held on July 26. That’s roughly 1/3 of the country, concentrated mostly in the north and in urban areas. Also, all the legislators facing recalls are KMT members who won in 2024, so this is almost all happening in places with more blue voters than green voters. We have never seen anything like this before, and it’s unlikely we’ll see it again. So what should we expect?

If you’ve been reading my blog for any length of time, you know my standard answer is “we don’t know.” And just like in previous posts before voting day, I’m going to argue that this time is especially unpredictable.

In most elections, turnout is a major unknown variable. We don’t know whether turn out will be in the low 60s or above 70, we don’t know which groups will turn out (especially young voters), and we don’t know which side will turn out its voters more effectively. Those questions all apply this time as well, except even more so. Turn out for recalls has varied from under 50% with both sides trying to over 40% with only one side trying. The recall of Han Kuo-yu was probably equivalent to 65 or 70% if both sides had tried, or more or less equivalent to what you would see in a general election. This recall will have a lot of national publicity, so I expect turnout to be fairly high. However, there are reasons to wonder. The DPP and KMT bases can both be expected to turn out, but we can’t be quite so sure about everyone else. Ko Wen-je pulled a lot of people to the polls in 2024, and he isn’t on the ballot this time. A lot of those voters might just decide that this isn’t their fight and they can sit this one out.

Both the KMT and the DPP have tricky considerations around how overtly to mobilize. The KMT has decided to try to mobilize its base, which makes a certain amount of sense. The recall threshold is low enough that they need to fight back, and, remember, most of these districts are fairly blue-leaning. However, there are probably not enough hard-core KMT true believers in most districts, and they need to win some districts that are closer to toss ups. If they scream frantically about green communism and DPP dictatorship, it’ll probably turn away as many people as they attract. Their best message is a much calmer reminder to voters that they want something to balance and constrain DPP power. Individually, the best message is constituency service. If the KMT goes all out screeching and howling, those calmer messages might get drowned out. Let’s also remember that the KMT polling numbers are not great right now, so those wavering voters sympathizers are already quite shaky.

The DPP doesn’t want to step on its toes of all the volunteers who worked so hard on the signature drives. It also would like this to be as large a coalition as possible and that might include some blue voters who don’t approve of the content passed by the legislature or the process used to do it. If the DPP stays in the background, those people can tell themselves they are voting against a specific legislator without expressing approval for the DPP. However, the DPP needs to ensure that competent campaigns are being waged everywhere. It has to make sure that the recall campaigns have adequate financing and advertising. And it also needs to make sure that there are some consistent themes about how rotten the KMT is.

*There are still 7 districts where the recall has not been finalized. For the purposes of this post, I’m guessing that New Taipei 11 and the two Nantou district school will all pass, but the other four (Taichung 2,3,8 and Hsinchu 2) will fail at the second stage.

I’m trying to make sense of the recalls by putting the districts into several buckets. The first way to look at recalls is through the most traditional lens, the partisan balance. The first bucket includes districts where the balance between the blue and green sides is fairly even. Most of these are districts that the DPP won in 2016 and 2020 (when their presidential candidate was pulling them to victory with 56 or 57% of the vote).

You might argue about the inclusion of one or two of these, but that’s nine solid targets that are realistic possibilities not just for recall but also for winning the by-election. This is why I have not dismissed the possibility of the DPP taking power in the legislature after this is all over.

The second bucket also uses a partisan lens, including districts that lean more blue but in which the green side has an outside shot.

These districts strike me as unlikely to support the recall and even more unlikely to change hands in the by-election if a recall were to succeed. However, they are close enough to the KMT can’t take them for granted.

The third bucket uses a completely different logic. This bucket contains controversial legislators. These are the people who have been at the center of all the controversial legislation and legislative tactics over the last year, and they are the ones who stoked the most anger and inspired the original recall drives. Many of them are in deep blue districts. For them it is not really a question of the party losing the seat, it is more a question of whether the voters want a new and better behaved KMT legislator.

This third bucket includes the three KMT floor leaders as well as several other legislators who are controversial for different reasons. To be honest, I can’t tell you why exactly I think the two from Taichung belong on this list; I just have a vague impression that they do. I surprised myself when I put together this list by leaving out several legislators who in the past would have been right at the top. But several of the former KMT floor leaders have been fairly quiet this term.

I don’t really expect anyone on this list (unless they were also on one of the previous two lists) to be recalled. That would be a seismic event that would force me to rethink many of my assumptions about Taiwanese electoral and legislative politics. But I will be paying attention to see if any of them seem to be paying a penalty for their behavior.

The 4^th bucket is completely different. This one is based on people who voted for Ko Wen-je for president. I do not consider these people to be monolithically TPP supporters. I think some of them are reliable TPP supporters, but many of them are wild cards who will vote for different parties or candidates in each election for a variety of different reasons that are difficult to summarize. More of Ko’s voters voted for a KMT legislative candidate than for a DPP legislative candidate, but there were plenty of districts in which that was not the case. More importantly, just because they voted that way last year doesn’t mean that they will vote the same way this year. This is especially true because there is only one candidate on the ballot in the recall. Remember, these are voters who don’t prefer to support either major party. It’s very hard to predict what they will do.

It’s also very difficult to know how they voted in 2024 in the legislative races in each district. Most surveys will give you a national estimate, but that doesn’t tell you anything useful about the race in Taipei 1, for example. However, I have an answer! It’s not a perfect answer, but then neither are surveys. At any rate, it’s the best answer that I know of.

My computer spent most of April and part of May running simulations to estimate how voters cast their ballots in the presidential and legislative elections. The central election Commission tells us down to the precinct level how they did this in each race, but it does not tell us how people who voted for a particular presidential candidate voted in the legislative election. I used an algorithm called multinomial ecological inference to estimate this. Let me explain using example of Taichung 5. This district has 298 precincts, and we know the exact results from each precinct for the presidential and legislative elections. This algorithm requires exactly the same number of voters in both elections, so I had to include invalid and indigenous votes in the model. In the legislative district there were 8 candidates, but six of them got fewer than 5% of the valid votes. Because this model isn’t good at estimating small numbers and each variable increases the computer time needed, I combined those six candidates together as “others.” This meant I had a 5×4 matrix, so I needed to estimate 20 values. Essentially, the algorithm starts from assuming votes are distributed equally and then starts drawing random numbers and adjusting the estimate a little bit each simulation. After many simulations, it will converge to a consistent result. How many is many? Let’s just say that running the model one time for Taichung 5 required over 72 billion random draws. What it eventually spits out is an estimate, not an exact measurement. If you did this all again, you would (hopefully) get something different but very similar. I repeated the process several times to ensure that all the results were telling basically the same story.

What story did the numbers tell about Taichung 5? Almost everyone who voted for Hou (95%) also voted for the KMT legislative candidate. An even higher percentage (97%) of people who voted for Lai voted for the DPP legislative candidate. Interestingly, when you compare these two numbers, the DPP candidate is leading by about 7000 votes. That’s not a huge advantage, but it is a clear margin. All he had to do was break even among Ko voters. That was not an impossible task. Nationally, Ko voters were more likely to vote for KMT candidates, but there was plenty of variation. There were lots of DPP candidates who managed this task. However, in this district the KMT candidate slaughtered the DPP candidate among Ko voters, winning this group by over 36,000 votes.

I believe that Ko voters are likely to be far less consistent from year to year than KMT or DPP voters and the TPP has undergone a remarkable degree of upheaval since the 2024 election, so this is where I’m looking for possible changes. The KMT won Taichung 5 by 27,000 votes, so it would take a pretty massive shift to overturn that result. Still, this is the kind of district I’m looking for in bucket #4.

Bucket #4 includes districts in which straight ticket DPP voters clearly outnumber straight ticket KMT voters and also in which the KMT’s superior performance among Ko voters enabled them to overcome this disadvantage and win the district. Most of these districts have already appeared in a previous bucket, but the point of this exercise is that they may also be vulnerable in a slightly different way. Also, I was surprised to see Taichung 5 on any of these lists, so I wanted to point that out.

The 5^th bucket is everyone else. I’m surprised by how few this is.

These are all deep blue districts that the KMT has no business losing. There might be some reason that one of these legislators has trouble, but it is an individual story rather than something systematic, at least as far as I can tell.

Overall, the legislators in bucket #1 are probably the most endangered. There are enough people in that category to suggest that this mass recall could cause a major political shift. However, one point of this post is to suggest that considerations other than pure partisan balance might also affect the outcome of the recalls. If several legislators from bucket #3, for example, lose or even come close to losing, we will have a ready explanation for what happened.

The deadlines for recall campaigns against 7 more DPP legislators have now come and gone, and none of them have been able to submit the requisite number of petitions. There are still 6 campaigns left, but none of them look very hopeful. The three yesterday and today, 2 in Taipei and one in New Taipei, were the critical ones. These are highly urbanized districts with a close partisan balance and should have been the easiest places to collect signatures and the most likely for the recall to actually succeed. They could not collect enough signatures in any of them.

Well, let me put a disclaimer on that. They claimed to have collected enough signatures in Taipei 1, but they were not allowed to file those signatures since the official head of the campaign did not show up. I suspect this was a face-saving measure. They showed up with about 20 boxes that they claimed contained more than 26,000 signatures. If you look at the pictures of other campaigns submitting their signatures, there are generally a lot more than 20 boxes. I suspect they knew that the head of the campaign wasn’t going to show up so they staged a photo op claiming they would have succeeded if not for his treachery. But I don’t think they had enough signatures in the first place.

Here are a few other pictures from campaigns that actually met the 10% threshold.

It isn’t expensive to put together a neatly arranged display of boxes with your campaign slogan printed on each one. Boxes are cheap. What it takes is time, energy, and a spark of creative joy.

This was a failure of the KMT to inspire and mobilize volunteers. There are plenty of partisan voters in the three districts with deadlines today and yesterday. The problem was that there weren’t armies of volunteers out on street corners and other public places collecting signatures and then taking them back and collating them.

A small personal anecdote highlights the difference. A few evenings ago, we saw four or five people out on a street corner holding signs asking people in traffic to recall our local KMT legislator. Keep in mind, this campaign has already collected and submitted plenty of signatures and will definitely pass the second stage. They weren’t asking for signatures. At the same time, no voting will take place for at least two months. And really, has anybody ever changed their mind about how to vote because they saw somebody on the street corner waving a sign? These people were on the street corner – in the rain – because they wanted to express themselves or they wanted to do something, anything, to help the recall. The KMT originally wanted to portray the people calling for recalls as radicals organized by the DPP. Well, there’s no competent Machiavellian organization in the world that would spend resources to mobilize 5 people to stand on the street corner two months before voting. Those people were entirely self-mobilized.

It is now highly likely that none of the recall attempts against DPP legislators will make it to the third stage.

This year’s recall vote will be an entirely one-sided affair. If you have an opportunity to vote, you will be voting on whether to recall a KMT legislator. The DPP will not have to play defense.