Anda di halaman 1dari 3

André Flykt aflykt@kth.

se

OPPONENT RECORD
Thesis compiled by
Andreas Serra, aserra@kth.se
Kai Widell Niigata, kaiwn@kth.se

Title of thesis:
Optimal Yahtzee performance in multi-player games

Opponent:
André Flykt, aflykt@kth.se

Was it easy to understand the underlying purpose of the project? Comments.


Yes, the abstract, title and introduction each separately described the purpose of the project
and all together it was even clearer. This is a good example of concise explanation of what the
project is about without getting too fuzzy about it.

Do you consider that the report title justly reflects the contents of the report?
“Optimal Yahtzee performance in multi-player games”
The title reflects the content of the report well. What did not come clear in the title was if the
report was aiming to discuss the subject or, as the authors did, implement an algorithm
searching for the optimal Yahtzee performance. The abstract did immediately straighten this
question mark, though.

How did the author describe the project background? Was there an introduction and
general survey of this area?
The introduction described the reasons why this topic is interesting and why the authors chose
it. The background covered the basic functions of Yahtzee, such as rules and how to score
points, followed by an explanation of the term “optimal strategy” which is essential for the
topic. The authors did not explain why they personally chose this topic.

To what degree did the author justify his/her choice of method of tackling the problem?
In the problem statement, the authors did narrow the topic down into three main questions.
Based on these I think the method description briefly discusses how they will proceed with the
questions. Sub-headings in the method define relevant definitions in a strict
programmatic/mathematical way. Also, the first sentence in the method refers to one of the
references, which is good. This gives the impression that the authors have done research and
chosen their method based on this.

Is the method adequately described?


The method takes up most of the essay and explains different sub-topics that will be crucial
for the project. Depending on how much technical knowledge the read has, it would be useful

1
André Flykt aflykt@kth.se

to explain terms like search space, algorithm, effectiveness of an algorithm etc. But since the
project revolves around programming I assume the authors assume that the readers are
coming from a technical background.

I am not sure if all the content is appropriate for the method though, some of the information
could be in the background. Information about what a single bot actually is, what AI are
etcetera fit better in the background in my opinion.

Has the author set out his/her results clearly and concisely?
The results take up a lot the essay’s space - which is good since it contains short summary of
how the implemented the solution and what the result was. The results are described shortly in
text - which could be a little bit more extensive - as well as in charts. The charts/images are a
good contrast to the rest of the rather strictly written essay.

There is a fine line between discussing and presenting results and I think some parts are more
to be considered as discussion/description of how they executed the project. For example how
states are stored.

Do you consider the author’s conclusions to be credible?


Since the bachelor thesis was about implementing their own algorithm, comparing it to
existing ones and draw conclusions from it, I think the conclusions and results are to be
considered as credible. One dimension that get lost is the code review though. Since no code
is attached, the reader has to rely totally on the authors’ description of the code. I think the
entire code should be attached, and short, critical sections should be put in the text and
described.

What is your opinion of the bibliography? What types of literature are included? Do you
feel they are relevant?
The reference list is extremely short, existing in only three references in total where one is
plain instructions how to play Yahtzee. I think this is the weakest part of the entire essay -
even though most of the project is about coding, at least a couple of references how you in
general attack these kind of computing would be in place.

Which sections of the report were difficult to understand?


The Method section contains strict, mathematical language and is appropriate but can be
difficult for some. Once again, it is important to have a clear understanding of what audience
is going to read the essay. A computer scientist will have no problems understanding, but I
would doubt a student from different faculty than math/computer science/technical
programmes would follow.

Other comments on the report and its structure.


The report is well formatted (in LaTeX) as well as the structure with headings. Introduction,
background, method, results, discussion and conclusion are the main headings and are to be
considered standard for essays like this.

What are the stronger features of the work/report?

2
André Flykt aflykt@kth.se

The authors kept the same strict language throughout the entire essay, and consistency are
good. The reader gets the feeling that the authors know what they are writing about since the
mathematical and probability terms are described with strict language and mathematical
definitions.

What are the weaker features of the work/report?


The authors assume a little bit too much that the reader knows about algorithms, probability,
perception of game strategy and math. I, as a computer scientist, follow but despite this I
would like to have some more explanations in order to remove any confusion.

Source code is not attached and I think it would be interesting to see parts of it. Not to dig into
how well written code the authors have produced, but to see some logic behind all the
probability formulas. Especially since the computations are quite

What is your estimation of the news value of the work?


When looking into the two references given by the authors, one quickly realize that this topic
has been covered before, which is reasonable. Looking at the results they are in the same
region as the other reports and I therefore don’t think that the news value are very high,
especially since you are not able to see the code at all. Low news value is not the same as
poorly executed project though.

Summarize the work in a few lines.


The authors have looked into the game of Yahtzee and by programming tried to implement a
optimal strategy for multi-player games and compare the algorithm against other. The
conclusion is that the single optimal strategy works well in multi-player games as well.

Questions to author:
1. How much input did you get when doing research about the topic? Is it a well
covered topic?

2. You mention that the only way for an algorithm to beat the single optimal strategy is
to take the opponents current score and available categories into consideration when
determine what to do. Could you give a brief explanation about this?

3. What was the hardest part about the project?