我特别拿ICNC课程中的下面这张图说明一下：

在各种抗议手段中，学者将它们分作三大类，第一类（第一行）是言语表达类；第二、三类是行动类。

如果看第一类的言语表达，你会发现，游行抗议示威都在这一类里面，换句话说，中国几十年来的所有的抗议中，即使最高境界的游行抗议，也只是言语表达，而不是行动，正因如此，中国不会有任何改变，因为行动才是最关键的。

如果看第二行，这是不合作的行动，包括罢工、抵制等。许多人可能以为这些都是无用的、也是软弱的，实际上，这些才是最常见的获胜方式。

总之，言语表达的抵抗如果没有行动的配合，那就不会成功。如果看过去二十年的公民维权，就可以看到，不仅没有进展，反而自709镇压后，萎缩到几乎没有了。

For English speakers, I’d still like to share with you what is said in the video, so I’ve transcribed and translated the video, and here’s the lecture in English（把上面的视频翻译成了英文，以便英文读者理解）：

Now let me briefly introduce decision tree, which is an important tool to be used to aid the sequential decision mode of the subject of political decision. Sequential decision refers to the decisions made by contestants in political struggles in a sequential way, which means one decision is made in response to the action initiated or taken by the opponent. So far as I know, the nonviolent conflict field of research has not yet introduced decision tree, but I feel it crucial for us, especially the parties in their very early stages of their causes. Decision tree theory involves four featured steps, namely, forward looking, inference, evaluation, and strategic decision on the equilibrium point.

Here are the meanings of each of the four concepts in a nutshell: forward looking refers to the expectation of the opponents’ options before you take any actions; inference means to think about all possible consequences that you could be faced with, which means you have to list out all of what you may think of regardless of their probabilities; evaluation refers to the process of giving points to both sides in question; then the last is to find the equilibrium in the contest, which is to inform your strategic decision.

Following these steps we make the decision trees. So in the next session, I’m moving on to illustrate how to draw on your knowledge and experience to map out a decision tree.

First of all, as the party initiating your action, you are the first mover. So here you draw a square, which is a token that you are to make your move. Next, write the name of your planned campaign in the square. Next decision is whether you are to make your campaign publicly known or develop by word of mouth, which is the key step. Now just for the purpose of this presentation, I am to change the title of the campaign into “Boycott Baidu”, which is not a real campaign but just for people to understand the session here.

Next comes to the government’s response to your campaign initiated. As it is known, to the publicized campaign, government is to respond in a way that may be different from the privately initiated one. Therefore, we draw two squares below the two pathways respectively, representing government under different conditions.

On the part of the opponent, in this case is the government, they have different options in response, and we are not sure which they are going to take. Therefore, we list out all the probable reactions from the opponent’s side under the this square. This necessarily means that there’s always a consequence of a responsive action from your opponent. Therefore, we draw a square under each probable consequence listed out in the map.

For each action, the opponent (or government in this case) has two responses, i.e., suppression and non suppression. Under suppression, it has three options, including legal proceedings with punishments such as criminalization and imprisonment; administrative penalties, such as fines, reprimands, invoking honors or certificates etc.; and extralegal means, especially the terrorist violence, such as hiring thugs roughing up the protester or assassination etc. Therefore, the opponents have four options in total.

After this, we are ready to take the stock of all the consequences that the campaign have to face with. For each probable response from the opponent (government in this case), there are four results: no change, which means the team maintains the same as before the campaign; expansion, which means the team becomes larger and stronger than before; shrinkage, which means the team becomes smaller and weaker; and dead, which means the team get dissolved as a result of the government’s response, but this doesn’t necessarily mean the team members are dead; it only means the team stops functioning and existing.

So under each line of opponent’s option, we draw a square representing the campaign and under each square of the campaign, we draw four probable results. We use circle to represent “remain unchanged”, large square for “expansion”, small square for “shrinkage”, and a diamond for “dead”. So now under the publicized campaign, we see 16 results, which we copy all and paste under the other pathway. Now we have 32 probable results in total.

So now it makes a complete decision tree. Next step is evaluation. We need to evaluate each result with marks. We assign two marks, writing two numbers in brackets under each result. The first number represents the first mover’s points; while the second, the respondent’s points. The points indicate each party’s degree of satisfaction on the result. I’m using a “0-10”point-scale here. But some would rather use “-5 - 5” point-scale. For me, a 10-point scale is clear enough to express the degree of satisfaction: a “0” mean the least of what you wished, while “10” means the most.

In this session, I can’t evaluate all the 32 results. I’m going to mark the eight scenarios of the “dead”. So now, we have finished step three. Now let’s turn back to the original map in the presentation, in which we see a note with an asterisk that says you can’t compare the results across different movers; you can only compare the different results under the same mover. Therefore, we can’t just say, under this result, we get 0 while the opponent gets 8, as compared with another result, where the opponent gets 10, this seems better. You can’t make such conclusions based on cross-actor comparisons.

Now after the completion of all the three steps, we are to decide on the equilibrium point. This depends on which is most probable and what you want as well. You can’t simply choose the highest scored outcome, which may not be the most probable or not likely to happen. In order to be well informed of each result, we need to use probability calculation to figure out the equilibrium point and strategize the planned campaign.

But before we move on to probability calculation, I’d like to use the decision tree to review a campaign that’s not premeditated or planned. I’m not going to draw the map but to use this ready-made map in the presentation. Reviewing on this map, you can see the the campaign is not the first mover. The movement happens mostly on an incident breaking out at the time, or the government’s new policy or introduction of a reformed policy or law, of which we have already seen many examples, such as an official’s death; a reformed policy on medical care; an introduction of an act, which was basically against people’s will, all of which were out of the agenda of the campaigns.

So in this map, what we see on top is the government’s actions, which means the government is the first mover. As a response, the campaign really doesn’t have many options. They either oppose or don’t. In the case that you don’t oppose, there’s no need to draw the results, which is also subject to the four possible endings. For instance, even if the campaign remains silent, your members may choose to leave or your social influence may be tarnished due to its silence. But if you oppose, you are also to choose whether you do it openly or secretly. Then consequently, you are subject to the same 32 results, just as what we have drawn in the previous map. However, the difference is that the results are not for you to choose, but subject to the government’s reactions toward you as the campaign, which means you do not have as much room to make the choice as in the previous map where you are the first mover, because you are just acting as a response, while the government has had responsive systems in place to act on any anticipated responses. So the results are more dependent on the government’s actions than of your own strategic choice, which means your ability to choose is weakened. That’s why we say in general, the campaign is in better position if it acts as the first mover.

(Will post English in a following note.)

關於calculated risk（計算內風險）：所附的視頻就是政治決策學科中專門講這類風險在政治博弈中其戰略決策中的關鍵作用。所有政治反對事業都是有風險的，最保險的是你不要去反對。

最近iyouport發的文章，我不能苟同，它分析編程隨想的被捕，講到他可能沒注意到數據以外的其他方面，特別舉例說，他採取匿名的另一面就是沒有一個可信任的人，所以，出事也沒人幫。

我對此深不以為然，我認為他是非常成功的一位反對者，他持續十二年，從概率上來說，若以一年為一個週期（這是運潮的典型週期），他等於是完成了十二個運潮，如此多人受益就是結果。而且報導說他在被捕前很久就感覺到異常，但他選擇不停止他的事業，也就可以證明他是有所準備的。

iyouport所說的讓某個可信的人掌握自己的信息，實際上更大的機會是他十二年也做不到就被逮捕。

反對事業是有風險的，不能尋求萬無一失，若真尋求萬無一失，那最好的是不要做，所以英文有一個詞，take calculated risk 也就是承擔計算內的風險。對編程隨想而言，以十二個週期計算，那是1/12的概率，所以，即使我自己，我也會去承擔這樣的風險，因為它已經高過許多的其他反對方式了。追求100%的保險，那就等於是不做。而認為必須找個人為被捕做準備，則一定會加大風險。

#[0]