随机的吧，中期偶尔会被迫研究原子理论

我也遇到过这个问题。难度是皇帝模式。都不敢碰最高难度。而且皇帝模式必须开作弊的全图否则估计打不过。

把必须的科技多留几个并行，只开必要的科技树，少开新科技树，比如除了识字你还需要多神论，之前选科技时就可以先研发骑术以打开多神论的研发路径，这样当下一个待选列表没有识字时，你还可以选多神论，总会有你需要的科技在待选列表中，前期就不会出现研发不必要科技的情况了。

根据外网的攻略，可研发的科技并不是每次都会出现在研发清单中
Every possible technology that we should be able to research will not be offered to us every time we have developed a new technology. The individual technologies will be offered following in this sequence: Yes, Yes, No, Yes, Yes, No, and so on. This should be perceived in the way that every tech will be offered twice in a row and then it will not be offered, then it will again be offered twice in a row and then it will not be offered and so on. The sequence will be moved one forward everytime a technology is found, achieved and traded. The sequence of technology offerings implies that it is dangerous to have Great Library because we will lose control over when technologies are achieved.
每个单独的可研发科技会以Yes,Yes,No的排序出现(Yes:出现，No:不出现)在研发清单中
即出现2次，接着不出现，之後再出现2次，接着不出现，一直持续下去。
每捡到新科技、自行研发、与对手交换科技，都会算一次排序。

楼上是台湾网友，哈哈