Archive for March, 2018

On polling primaries

March 8, 2018

The KMT and DPP have now finished their first round of polling primaries. It seems to me that the polling primary might not be around very much longer. But before I get to that, let me start this post with some quick comments about the DPP results of the past couple days.

We have been headed for factional wars in Kaohsiung, Tainan, and Chiayi for a few years now, but it turned out that the full-blown battle only materialized in Chiayi. Chen Ming-wen’s 陳明文 side has won that battle, so he and his friends will control the county government for the next four years. This will make a full two decades in power for him since he switched sides in 2002. To recap, during the 1990s Chiayi County was reliably KMT, with the DPP never really coming close to victory. Even in the 1997 DPP tidal wave, Chiayi stayed blue. Chen himself was the second most important politician in the Lin Faction, which was clearly the second faction in the KMT. The dominant Huang faction won the intra-party struggles, and then the KMT won the inter-party struggles. Two things changed this. First, the dominant Lin faction politician, legislator Tseng Chen-nung 曾振農 (famed as king of rattan chairs涼椅王), faded away. I can’t remember clearly, but I think he had some legal and financial problems and eventually went to exile in Cambodia where he died in the early 2000s. Chen Ming-wen thus assumed leadership of the faction. Second, Chen read the changing political environment, and showing considerable political skill convinced nearly his entire to switch sides. Inside the DPP, Chen’s faction was the dominant force, and the newly strengthened DPP easily beat the KMT. Chen served as county magistrate for two terms and then turned the county over to his ally (and Tseng Chen-nung’s widow) Helen Chang 張花冠. During her two terms she apparently decided to try to take over leadership of the faction, and so here we are today. The Chen and Chang factions went head to head in the primary, and Chen’s candidate Weng Chang-liang 翁章梁 won a narrow but clear victory 43-35. I really only know three things about Weng: 1) He is in Chen’s faction, 2) He comes from the Wild Lily student movement, and 3) He has three surnames 翁、章、梁.

The Kaohsiung primary momentarily turned nasty about a month ago, when mayor Chen Chu 陳菊 released her book and rehashed some of the unpleasantness of her 2006 campaign. It turned out to be a terrible strategy, and her favored candidate Liu Shi-fang 劉世芳 ended up withdrawing from the race. This left the field wide open for Chen Chi-mai 陳其邁, who was probably always going to win anyway. However, without Liu (and with mayor Chen staying neutral), Chen Chi-mai lapped the field, getting more support than the other three candidates combined. This large win makes it highly unlikely that anyone from the green side will challenge him in the general election. Unless something really strange happens, Chen is a shoe-in to win in Octobler.

Chen Chi-mai’s smashing victory in the Kaoshiung primary is important for another reason. Chen has a good head on his shoulders and plenty of ambition. He probably slots into third place in the future DPP presidential candidate sweepstakes, just behind Premier William Lai 賴清德 and Taichung mayor Lin Chia-lung 林佳龍. I think he jumps ahead of other contestants, such as VP Chen Chien-jen 陳建仁, Taoyuan mayor Cheng Wen-tsan 鄭文燦, Chen Chu, and people not currently in the DPP such as Huang Kuo-chang 黃國昌 and Ko Wen-je 柯文哲. I have a very high opinion of Chen Chi-mai, and I expect his national profile to soar over the next few years.

The contest in Tainan was also a blowout, with Huang Wei-che 黃偉哲 clearly beating Chen Ting-fei 陳亭妃 and the rest of the field. Tainan can be a launching pad to the presidency (see Lai, William), but I don’t have such a high opinion of Huang. Thus far in his career, he has been better at going on talk shows than actually doing politics. But there is absolutely no chance that the KMT will win Tainan, so Huang will get a chance to prove himself and make me eat my words.


I don’t really want to write about the specific races. I want to write about the future of the polling primary. I think this institution might be on its last legs.

Nominations are always challenging. Parties want to nominate a (1) candidate who will win the race and is (2) ideologically consistent with the party’s mainstream values. They further want to do this in some sort of process that (3) gives the party some degree of control over who they nominate, and (4) convinces the other aspirants not to challenge the decision in the general election. These various goals usually conflict with each other. In the late 1980s and early 1990s, when elections were becoming more and more competitive, both parties struggled with nominations. The KMT tried decentralizing its process, but it ended up with candidates the party leaders didn’t like. It also found that candidates who were highly rated by the local party office weren’t always great candidates, such as the 1989 Taipei County magistrate debacle in which all the local factions rebelled against the bookish and clueless Lee Hsi-kun 李錫錕. (Somehow, Lee has re-emerged after all these years to launch an internet-fueled campaign for Taipei mayor. The world is weird.)

The early KMT “primaries” were generally advisory and non-binding. The DPP tried holding binding votes among party members. This resulted in widespread accusations of mindless factional vote trading, cultivating phantom party members, and outright vote-buying. When the DPP tried giving more weight to elite party members’ preferences, the factionalism only intensified.

This clearly didn’t work. The DPP tried moving toward an American-style public primary in the 1996 presidential nomination. However, in the USA, the primary elections are run by the state. The Taiwan government was not interested in helping the DPP run its primary election. Instead, the DPP launched a touring roadshow. They set up a stage, just like a political rally, and the two finalists debated each other (saying the same things every night). After (ok, during) the debate, the audience could go to the side tent and vote by casting a special commemorative coin into a custom-made vending machine. It was a disaster. The machines broke repeatedly, there were rumors of repeat voters and other forms of cheating, turnout was much lower than the DPP had hoped for, and the DPP fundamentalists ended up nominating Peng Ming-min 彭明敏, one of the worst candidates in Taiwan’s electoral history. (He could barely be bothered with unimportant things like, you know, policy.) The DPP was never going to win in 1996, but it was clear that in a more winnable race this nomination process might torpedo their chances.

Telephone polls were the answer. At first, parties used a mixture of party member voting and telephone polls. However, the polls were seen as a more authentic gauge of support in the electorate. A candidate who lost the party vote but won the telephone poll was not going to withdraw in favor of a candidate who had won the party vote but lost the telephone poll. Moreover, as long as there was some component of party voting, there was a danger of vote buying. The death knell for party voting came prior to the 2008 election, when the electoral law was revised extending the penalties for vote buying to primary elections. The parties (especially the DPP) were terrified that their nominees would be indicted for vote buying BEFORE the election. Since 2008, the parties have resolved all of their nomination disputes by polling primaries. They don’t always need the polling primary; sometimes they can negotiate a nominee acceptable to everyone. However, the polling primary is always the default final solution.

Telephone polls were particularly useful because they convinced losers to accept losing. When they work properly, polls give each candidate a fair chance to demonstrate their support. It is hard to argue that you are actually the more deserving nominee when a telephone poll with a representative sample says the opposite. Polls are also hard to manipulate, especially in comparison to other alternatives such as voting by party members. Because each resident with a telephone is potentially a polling respondent, it is inefficient to buy votes or intimidate voters.

To top it off, polling is pretty cheap, so parties can easily afford to hire multiple companies and do large samples.

So what is the problem? I think that polls are losing their authoritativeness. For one thing, there are rumors that innovative politicians are finally learning how to manipulate polls. We hear again and again that politicians are buying thousands of telephone lines to increase the chances that pollsters will call them. I have doubts about this. Wouldn’t Chunghua Telecom notice a person registering 1000 new numbers at one address and flag that transaction? Who would staff all those telephones? Besides, you need a lot more than a thousand extra numbers to move the needle in, for example, Kaoshiung, which has over 2 million residents. Nonetheless, rumors of manipulation are an important ingredient.

The second ingredient is real. There is a growing crisis in polling caused by the rise of mobile phones. For decades, pollsters have assumed that a representative sample of household landlines was effectively a representative sample of the entire population. Every house had a phone, so a sample of phone numbers covered everyone. This was never actually true, but it was close enough. No longer. Now we don’t quite know how construct a representative sample. It is clear that you cannot rely simply on landlines. However, there is no theoretically rigorous algorithm to combine landlines, cell phones, and internet surveys. That 3% margin of error (which is calculated solely by sample size) you see reported in every newspaper story was always something of a lie, because a sample of landlines was never actually a true random sample. Now it is even more meaningless.

Note that I did not say that pollsters are intentionally manipulating the results. Professional pollsters don’t do that, especially for something like a party primary. The protocols are set up to allow observers from each candidate to make sure that no manipulation occurs during the interview process.

Many candidates don’t know what they are looking at; the speakers of the Keelung and Chiayi councils are not exactly members of the numerati. They might really think they have somehow been cheated because, for instance, the polling company refused to call the list of telephone numbers they brought in. Other candidates, such as Chou Hsi-wei 周錫偉 in New Taipei, probably do understand the basics of polling but cynically proclaim doubts that the results were manipulated to explain away their losses. The problem is that when these sorts of accusations are made (as they always are), the experts no longer fight back with a unified front. We no longer have absolute confidence in our own results. Rather than a solid wall of rejection from the experts, the complainers now have little cracks to exploit. Polls simply don’t have the authority that they did ten or twenty years ago.

Without this aura of authority, polling primaries will probably start to fail at one of their most basic tasks: convincing losers to accept defeat. That means that parties will need a new mechanism. I suspect the next thing they will try might be an American-style publicly run primary election. However, that will require significant legal changes, so it is still some distance in the future.