Reducing latency in presenting digital videos

A method for reducing latency in presenting a video and performed by data processing apparatus, the method comprising:

• selecting a video in response to a first content request received from a given client device;
• providing, to the given client device, at least a portion of the selected video;
• determining that less than all of the video was presented at the client device, the portion being less than a complete duration of the video;
• in response to determining that less than all of the video was presented at the given client device:
• reducing latency in subsequently presenting additional portions of the video at the given client device by storing the video in a local cache for the client device;
• and
• adjusting distribution criteria for the video for subsequent content item requests received from the given client device based on how much of the video was already presented at the given client device, including adjusting the distribution criteria such that an increase in how much of the video has already been presented at the given client device results in an increase in a likelihood that the video will be selected for presentation at the given client device in response to subsequent content requests received from the given client device;
• receiving a second content request from the given client device;
• and
• causing the selected video to be streamed from the local cache and presented at the given client device in response to receiving the second content request.

A non-transitory computer storage medium encoded with a computer program, the program comprising instructions that when executed by one or more data processing apparatus cause the data processing apparatus to perform operations comprising:

• selecting a video in response to a first content request received from a given client device;
• providing, to the given client device, at least a portion of the selected video;
• determining that less than all of the video was presented at the client device, the portion being less than a complete duration of the video;
• in response to determining that less than all of the video was presented at the given client device:
• reducing latency in subsequently presenting additional portions of the video at the given client device by storing the video in a local cache for the client device;
• and
• adjusting distribution criteria for the video for subsequent content item requests received from the given client device based on how much of the video was already presented at the given client device, including adjusting the distribution criteria such that an increase in how much of the video has already been presented at the given client device results in an increase in a likelihood that the video will be selected for presentation at the given client device in response to subsequent content requests received from the given client device;
• receiving a second content request from the given client device;
• and
• causing the selected video to be streamed from the local cache and presented at the given client device in response to receiving the second content request.