Sources

The short version

• Everything below is public. Nothing in this course rests on a private file or an unnamed source.
• The tables are generated from the recovery's own manifest, so they cannot drift from what we hold.
• Rows in grey are the holes: files the archive reached only after they were already gone.
• If a link here dies, tell us and we will repoint it. That is rather the point of this project.

A project that exists because a website died should be able to say exactly where every one of its own claims came from. So here is all of it, grouped by where it came from, with the snapshot each file was taken out of.

Rows in the manifest

0

Recordings

5, all on YouTube, none rehosted here

Replays recovered

13

Problem sets recovered

4 of 6

Files the archive missed

0

The five recordings, on YouTube

One row per lecture, under the class's own week number. The manifest also holds the metadata and description files the research pulled alongside each video, which are not listed here because they are our working copies rather than sources in their own right. We never rehost the video.

What we publish, and what we do not

The manifest is a list of what the research holds. It is not a list of what this site serves.

The five lecture recordings stay on YouTube. We embed them from youtube-nocookie.com, only after you press play, and we never rehost a copy. The transcripts we made are a private research instrument: they make the tape searchable and quotable to the second, and they never reach a browser. What you read on the week pages is our own notes.

The archived documents are linked, not reproduced. Homework is re-posed in our words with the original one click away. The syllabus is presented as facts in our own layout. Sirlin's write ups are summarised, credited and linked, never republished.

The thirteen replays are the exception, because a replay is a record of a game rather than a piece of writing, and because playing one in the engine is the only way to make it mean anything. They are served by slug from a whitelist and the handles you see are the ones in the file.

What is lost

This is the part most projects leave out. Here is what we looked for and did not find.

• Homework 2, 3, 5 and 6

  The PDFs were captured only in 2011, after the domain lapsed, and every one comes back as a one byte stub. Sets 2 and 3 are recoverable anyway, because their solution keys restate each question before answering it. Sets 5 and 6 are gone.

• The reader

  The Theory of Starcraft by Alan Feng, sold on paper at a copy shop near campus in two editions. No copy is online. This is the single biggest hole in the record, because it is the text the lectures were built to accompany.

• The slide decks

  Never posted. Asked for in week 1 and refused on the night. The only versions of those slides that exist are the frames of them in the recordings.

• GamePro's copy of week 1

  The class site pointed at a zip file on GamePro's CDN. That CDN has no captures at all in the Wayback Machine, so the professional recording exists only as the YouTube upload made from it.

• The blip.tv copy of week 3

  Academic Earth hosted its own video file rather than embedding YouTube. It would have been a better copy than the 320 by 240 upload. blip.tv shut down in 2015 and no capture of that file exists.

• Week 12, entirely

  No recording, no write up, no note on the class site, nothing. It is the one session of the fourteen with no witness of any kind.

Where we looked

Citing us

Every page carries its own sources block with exact URLs. If you are citing the course itself rather than our pages, cite the class: Feng, Alan, Chen, Anthony and Mahbod, Sherwin, Game Theory with Application to Starcraft, UC Berkeley DeCal, spring 2009, and then the archived syllabus.

If we have something wrong, the errata page is where the correction goes, with your credit on it.