Interval Licensing LLC v. AOL, Inc. et al

Filing 241

PRAECIPE re (240 in 2:10-cv-01385-MJP, 240 in 2:10-cv-01385-MJP, 240 in 2:10-cv-01385-MJP, 240 in 2:10-cv-01385-MJP, 240 in 2:10-cv-01385-MJP, 240 in 2:10-cv-01385-MJP, 240 in 2:10-cv-01385-MJP, 240 in 2:10-cv-01385-MJP, 240 in 2:10-cv-01385-MJP, 240 in 2:10-cv-01385-MJP, 240 in 2:10-cv-01385-MJP) Statement,,,,,,,,,,,,,,,,,,,,, by Plaintiff Interval Licensing LLC. (Attachments: #1 Exhibit 1, #2 Exhibit 2, #3 Exhibit 3, #4 Exhibit A-1, #5 Exhibit B-1, #6 Exhibit C-1, #7 Exhibit D-1)(Berry, Matthew)

Exhibit B-1 (Amended)

ATTORNEY DOCKET NO.

INT1P206+

IN THE U.S. PATENT AND TRADEMARK OFFICE
Provisional Application Cover Sheet

ASSISTANT COMMISSIONER FOR PATENTS
Washington, D.C. 20231
Sir:

This

IS

a request for filing a PROVISIONAL APPLICATION under 37 CFR 1.53 (b )(2).

g

INVENTOR(s)JAPPLICANT(s)

Residence (City and Either State or Foreign Country) ti..~

FIrst Name, MI

Last Name
Naimark
Bergman
Wed

rJi~ - c c
~
at---

Michael
Aviv
Emily

Moresco

G

Ignazio

;:).,...... -

C\:

T"l _ _- ,

;;;~

--------------------~-------T-IT-L-E--O-F-T-H-E-I-N-V-E~N-T-I-O-~---------------------------;~

G

ALERTING USERS TO WEB SITES OF CURRENT INTEREST AND HANDLING LARGE INCREASES IN USER
TRAFFIC

CORRESPONDENCE ADDRESS

Customer No. 21912
RITTER, VAN PELT & YI LLP
4906 El Camino Real, SUite 205
Los Altos, CA 94022
ENCLOSED APPLICATION PARTS (check all that apply)
( X)
( X)
( )
( )

Specification
Number ofPages .2...
( X) Small Entity Statement
Drawing(s)
Number of Pages 2
Power of Attorney
Additional inventors are being named on separately numbered sheets attached hereto.

METHOD OF PAYMENT

f""

A check in the amount of$

150.00 to cover the filing fee is enclosed.

At any time during the pendency of this application, please charge any fees required or credit any overpayment to
Deposit Account No. 50-0685 (Order No. INTIP206+).

Express Mail Label No EL422310950US
Date of Deposit

January 28, 2000

I hereby certIfy that this IS bemg deposIted With the Umted States
Postal Service 'Express Mal1 Post Office to Addressee' service under
37 CFR 1 . 1 0 on the date mdlcated above and is addressed to the
Assistant Commlssloner for Patents, Washmgton, D.C 20231.

Lee~Pelt

Attorney for App1icant(s)
Reg. No. 38352
Date: January 28, 2000
Telephone No.: 650-903-3500

~*
Typ~lmper

By

,

IL DEFTS0009124

Alerting Users to 'Web Sites of Current Interest and Handling Large Increases in
User Traffic

by Inventors
Michael N aim ark
Aviv Bergman
Emily Weil
Ignazio Moresco

SUMMARY
Web cameras and web video are undergoing explosive growth, due in part to broader
bandwidth, better compression technologies, and cheaper cameras. One can liken the
Web to "million channel television." However, most of the cameras show nothing of
interest to most of the watchers most ofthe time. While dozens of web cam portals and
directories exist, none are capable of propagating an alert that "something interesting is
happening now," to the right people. To solve this problem, a real time meta-data
infrastructure allowing people who see interesting occurrences to alert other interested
parties is disclosed. The system is referred to as "Hot Now." People who receive an alert
may further propagate the alert to broader and broader audiences, causing a swarm of
users to visit the hot site. A method of preventing server overload when such "mass
swarming" occurs is also disclosed, as well as a strategy for caching and archiving the
selected video segments. In addition, other examples of "Hot Now" applications in
addition to webcams are suggested.

1. Background
1.1. Web cams and Web Video
The first "webcam" appeared in 1991 (actually before the World Wide Web) in the
Trojan Coffee Room at the University of Cambridge, for members of the Computer Lab
to see how much coffee was left in the coffee pot. By 1996, approximately 100 live web
cameras existed. By July 1999, web cameras were being bought worldwide at a rate of
over 1,000 per day. Therefore, the current number of web cams may exceed 100,000.
This should come as no surprise to anyone monitoring trends in web and video
technology. As modems get faster and broadband technologies such as cable modems and
DSL come into use, as video compression allows higher-quality video to be efficiently
sent and received, and as camera costs decline, one can easily conclude that web cameras
will continue to proliferate. It may appear overly dramatic today to consider the Web as
containing "million channel television," but such claims will likely be realized in the nottoo-distant future.

1
IL DEFTS0009125

A similar revolution is already in progress in the field of video production. Over the past
decade, the means for logging, editing, mixing, and adding special effects to video has
quietly moved from expensive post-production facilities to the desktop. For the cost of a
day's use of such facilities, home videographers can now own a video camera, computer,
and video editing software. What has occurred with word processing and spreadsheets
has now occurred with motion picture production.
Given the proliferation of live webcams and tools for video production, the bottleneck
now is distribution and access-- finding the content the user is looking for and making
sense of the data.

1.2 Liveness, Freshness, and the Shared Viewing Experience
A simple solution to distribution is having web video downloaded by a user requesting a
particular selection from a website. This is a good and obvious approach for precomposed video. But an additional element exists for live and near-live ("fresh") video
from webcams, and for pre-composed video webcast to anyone interested: a shared
common viewing experience.
In many instances, there is value in such a common experience. For example, gossip
about last night's favorite television show around the workplace water cooler, the
popularity of live televised sporting events, and the "did-you-see-that?" discussions
around rare webcam events. The recent explosion of real time "chat" or "instant
messaging" on the Web further suggests a strong desire for live real time shared
expenences.
There also is value in near-liveness or "freshness." Consider the difference between
seeing a live webcam image of a rhinoceros at an African watering hole as it happens
versus finding a stock video of a rhino. Now consider coming home from work and
making a Web query "show me 'sightings' that happened today at the African watering
hole?" Such "freshness" is close to liveness, and both freshness and liveness have value
distinctively different from "canned" video. Freshness value is driven primarily by the
cost of caching compared to its demand. It is expected that freshness has a significant
positive value for at least 24 hours.

1.3 Web Video Portals, Directories, and Rings
Not far behind the proliferation of web cams and web video is the proliferation of web
services to help people find particular topics. These web cam directories or "portals"
essentially mimic non-video portals by consisting of hierarchically organized keyword
searches (e.g.,finches: birds: animals: general interest). Keywords are determined for
video by humans since computer vision is not currently able to automatically recognize
the contents of an arbitrary video image. Today, dozens of such webcam directories exist,
some including more than 10,000 entries. Such services are valuable in a limited way.
They can help users find the African watering hole, but cannot help users detennine when
an animal is present.

IL DEFTS0009126

A variant of webcam portals and directories noteworthy due to their popularity are
web cam "rings." A ring is a group of web cam operators who share a common interest,
whether it's animals, landscapes, or nudes, and offer the service of allowing users to
move "around the ring" with controls like "next" and "back." Though rings are generally
open to new members, they have the feel of hobbyists , networks (indeed, their
functionality breaks down if they get too large). While rings suggest an alternative to
portals and directories (which one may predict will be short-lived), they also suggest an
extraordinary enthusiasm among participants.
1.4 Voting and Polling

Most webcam and web video directories have some method of ranking. These methods
range from editorial choices made by the directory operators to voting on the part of the
viewers. It's common to see "top ten" lists, often with voting numbers available, and to
see such honors as "webcam of the day." From our perspective, such determinations are
relatively static and cannot help anyone interested in short time based events. Sites
which list a webcam of the minute do exist, but there is no special time-based relevance
in a selected webcam.

2. Hot Now ("Bitswarm")
2.1 What It Is

Hot Now is based around a unique meta-data infrastructure that allows people who are
first to see an interesting web video event to propagate an alert to others who may find
the event interesting, and to do it as fast as the Internet will allow. This concept is also
referred to as "bitswarm." Bitswarm uses active human participation and the power of
distributed human intelligence.
In one embodiment, a "Hot Now" virtual pushbutton is present on a user's web display.
When the user sees something they feel is of interest, they press the button. Pressing the
Hot Now button sends an alert message to everyone using the infrastructure who has
indicated that such alerts are of interest to them (based upon factors described below).
Along with the alert message a link to the website of interest is provided, and alerted
users can chose to go there. If they also believe the site is currently interesting, they can
press their Hot Now button and further propagate the alert.
It is not required that everyone press the Hot Now button when they believe that what
they are watching is Hot Now. So long as a proportion of the alerted community acts,
propagation will occur. A simple Hot Now button interface encourages more
participation.

While the Hot Now infrastructure uses human recognition and human decision-making, it
may also be augmented by machine recognition and intelligence. For example, simple

.3
IL DEFTS0009127

motion detection can be used to send someone to investigate the African watering hole if
a motion threshold is exceeded. Further propagation of the alert depends on humans
deciding whether or not to press the Hot Now button.

2.2 Propagation Rather than Polling
The disclosed method of propagation is superior to polling. Polling is one-shot and
generally static, while propagation is multi-step and dynamic. Propagation builds on what
already exists, from a single alert which may alert 100 people who may then each alert
another 100 people, and so on. As more people propagate an alert, more people are
alerted. As such, propagation can produce exponential changes occurring in short periods
of time, classic positive feedback behavior.
It is equally important to understand that propagation can have a negative value as well as
a positive one. By not pressing the Hot Now button, an alert will decrease in strength due
to decay. Propagation both positively propagates something interesting and negatively
propagates (filters) something that is not interesting or is no longer interesting. Such
propagation is therefore a closed-loop self regulating system.

In some embodiments, a "not Hot Now" button is also provided. Also, a scaled Hot Now
button, e.g., from -10 to +10, further amplifying the alert may be provided. In general, a
tradeoff exists between complexity and motivation, and user behavior is kept as simple as
practical.

2.3 Factors and Specifications
2.3.1 Hot Now Input
The Hot Now interface consists of an alert button and a text field. The alert button and
text field can be integrated directly into the content of a web page, much like a banner ad,
or incorporated into a small floating browser window.
When an alert is triggered, two values are transmitted to the server: the URL being
watched and the alerter ID. A user can also opt to send a text comment.
Each client application may monitor the frequency of alerts. Abusers of its functionality
can have their alert access restricted; productive users can have their alert access
increased.

2.3.2 Hot Now Propagation
A user receives an alert if she is interested in a) the alerter' s interest or b) the URL' s
content category. Interest may be expressed by setting filter variables. The filtering
interface is described below.

IL DEFTS0009128

Interest groups and URL categories are hierarchical. For example, the "Bird Watching"
interest group is a subset of the "Animal Lovers" group. When enough "Bird Watching"
members trigger a Hot Now, the alert passes up the hierarchy to members of the "Animal
Lovers" group.
As more members trigger the Hot Now, the system can detect overlapping of clicker
interests. For example, when a sports enthusiastlanimallover and a sports
enthusiast/news buff both hit the Hot Now for the same URL, the system primarily alerts
sports enthusiasts.
2.3.3 Settings For Filtering Output
Users control the influx of alert calls by selecting the following:
•

•
•
•

"

Interest Group Bias: increases a client's sensitivity to alerts triggered by
members of specific interest groups (e.g., birds, animals, weather, natural
disasters, car crashes, sex).
Clicker Biases: heightens sensitivity to alerts from specific members of the
community (e.g., registered club members, democrats, women).
URL Biases: favors alerts associated with particular URLs or URL categories
(cameras located in South America, cameras set up by National Geographic).
Heat Threshold has two components: "heat sensitivity" determines the number
of alerts required to announce an event to the user; "cooling" determines the
duration after which an event will no longer be announced to the user.
The "heat sensitivity" variable lets a user favor particular stages of a "Hot Now"
event. At one extreme, "heat sensitivity" senses URLs that have received only a
single alert. At the other extreme, "heat sensitivity" senses only the hottest URLs,
i.e. URLs that have received many alerts. This setting can be thought of as
ranging from "I'm so interested in this that I want to be alerted first (even if I
have to deal with false alarms)" to "don't bother me unless many people already
find this hot."
The "cooling" variable is used to calculate relative heat ofURLs. The variable is
segmented into intervals oftime. An alert during the most recent "cooling"
interval has a greater heat value than an alert during the least recent interval.

•
•

Hot Now display: controls the number ofURLs displayed.
Comment Flag: controls the display of user comments accompanying alerts.

Other filters are set automatically:
• After a user visits a hot site, she or he temporarily becomes less sensitive to alerts
from that site.
• An event's first alert is "hotter" than subsequent alerts to the same event.

5
IL DEFTS0009129

Either the server or the client can filter alerts. In one embodiment, the server updates each
user's settings in a database of user profiles and transmits a pre-customized Hot Now list
to the client. In another embodiment, the client customizes raw data received from the
server; and settings are updated on the client and saved to the server at the end of each
seSSIOn.

2.3.4 Hot Now Output Display
Alerts can be displayed, depending on bandwidth, in the following formats:
• Lists ofURLs
• Thumbnails of a web page
• A single URL's "heat" display
• Animations
• Other visualizations

It is also possible for an alert to trigger external devices using different modalities than a
standard computer, such as a pager, telephone, or lights flashing.

2.3.5 An Example Hot Now Architecture Specification
In one embodiment, URL and alerter interest groups are the same, based on a standard
list of topics (which mayor may not be hierarchical). Each user selects a series of interest
groups and sets a sensitivity threshold for each selected group.
Preferably, interest group filtering is implemented on the server and sensitivity filtering is
implemented on the client.
Alert Messages Sent To the Server
Data: [URL, AlerterID, Comment]
The comment variable is optional and may either be an open text field or a pull down
window with pre-assigned comments such as topics.
Alert Message Sent from the Server to Users
Data: [List ofInferred Interest Groups, URL, Comment]
Each alert is propagated to members of a hot event's inferred interest groups. The
inferred interest groups include members of the URL's interest group and overlapping
alerter interest groups.

IL DEFTS0009130

The Inferred Interest Group Process
If there is no overlap of interest groups, the server sends alert messages to all members of
each alerter's interest group. If there is overlap, the server sends alert messages to those
within the intersection of interest groups.
Repeat for every alert
If a region of interest overlap is not reaffirmed by an incoming alert, it looses importance.
Overlapping regions may shift over time.
Server Processing
For each alert received, the server performs the following:
•
•
•
•

Looks up the alerter's UserID for her Interest group selections
Looks up URL Interest groupings
Performs inferred interest group algorithm
Searches for UserID's in inferred interest groups and transmits a message

Sensitivity Filtering on the Client
A user's sensitivity selections are saved on the client in the following table:
[Interest Group, Sensitivity Threshold, Timespan]
Timespan is the length oftime during which a URL's alerts are counted. A URL is
displayed if its sum of alerts reaches its threshold before its timespan has expired.
At the end of each timespan, the URL's count is set to zero. Timespan is initially a
default value that can be reset by the user.
A dynamic table keeps track of the count for each hot URL:
[URL, Interest Groups List, Counter Time List [Tl, T2, T3 ... ] ]
A URL is displayed once its lowest interest group threshold is reached.
Credible Alerters
The system can recognize a first alerter and can keep track of responses to her initial
alert. An alerter gains credibility when her alert attracts many responses. A credible
alerter's alert is propagated with greater magnitude than a non-credible alerter. A credible
alerter's alert is sent more than once to all her interest groups (regardless of inferred
interest groupings).

7
IL DEFTS0009131

The credit system creates leaders. Leaders create other leaders. A credible alerter might
respond to the call of a non-credible first alerter. If she sends a credible alert, enough
users will probably respond so that the first alerter will become credible.
Credibility within the community shifts and decays over time.

3. Overload Protection ("Bits urge")
The Hot Now meta-data infrastructure may potentially crash webcam servers at the
moment when the most interesting video is occurring, due to massive
herdinglflocking/swanning by alerted users. To solve this problem, an overload
protection service referred to as "Bitsurge" is implemented.
Bitsurge monitors alerted servers for overload. If an overload is imminent, the
overloading web page is copied to a larger Bitsurge server and traffic is automatically
rerouted to the Bitsurge server in a manner that is transparent to alerted users. The
Bitsurge server becomes the invisible intennediary.
When a web producer registers her site into the Hot Now network, she downloads a
Bitsurge application and installs it onto her server. As users flock to her site, Bitsurge
sends the site's data to the Hot Now server. Each client request for the site is then
redirected from the original server to the Hot Now server.
Bitsurge caching persists during the span of a Hot Now event. By keeping track of alert
frequency for each site, the Hot Now system can detect a site's Hot Now event before the
site's original server is overloaded. As soon as a site receives many alerts, the system
assumes that a flock is on its way.
Alternatively, a Hot Now event can be detennined beyond the Hot Now network, by a hit
counter running on the site's server. When a site is hit by many users, the Bitsurge
application detects that a Hot Now event is occurring - and redirects data to the Hot Now
server.
Bitsurge has applications independent of a Hot Now alert. Any server that may
experience overload may benefit from such a service, particularly when the overload is
occasional or unpredictable.

4. Caching and Archiving
As mentioned earlier, near-live or fresh content has value similar (and in some cases
greater) than live alerts. For people to see what was hot that day (or some other short
period of time), real time caching of Hot Now alerts is used.
What to cache and when to cache alerted videos is partly market-driven and partly
context-driven. Alerted events that are known to have short durations (e.g., celebrity

IL DEFTS0009132

sightings in public places) require recording to begin almost as soon as the first alert
occurs, while events that have longer durations have looser constraints (e.g., if a rhino
stays an average of 10 minutes at the watering hole).
Caching for freshness is, by definition, temporary. Ifthe goal is to provide a commercial
service, at some point the value of the material drops below the cost of caching it, as its
freshness turns "stale." Hence, a fresh cache may be regularly flushed.
A symbiotic relationship exists between cache flushing and archiving. The goal of
archiving is to save "the best" from a sample far too large to archive in its entirety, and
flushing the cached material to an archive on a regular basis benefits all parties. The
webcam operator may have a minute of his or her material found through the Hot Now
alert infrastructure, made available while its fresh (for fame or fortune), and then made
part of the permanent collection of an archive. The Hot Now system benefits by finding
and caching what's Hot Now. And the archive gets the best of the best, as determined by
a "people's choice."

5. Other Applications
The Hot Now "bitswarm" system, as well as the "Bitsurge" overload protection, has
applications beyond webcams and web video. It has value for any networked phenomena
that changes quickly.
One class of applications also involve the Web. For example, the system may be used to
provide and alert when someone finds anything on the Web that is timely and worthy of
alerting others who have expressed interest, such as auctions.
Another class of applications are non-Web networks. For example, a broadband
television environment with several hundred channels and a simple Hot Now
infrastructure may be used to help users select channels. For example, a Hot Now button
on a remote control with 4 categories to select (e.g. nUdity, funny moments, news flashes,
and sports climaxes) and only 1 hierarchical level (top level is general interest) may be
implemented. Hot Now alerts are propagated when the Hot Now button on the remote is
activated during a program. Given how many people already are "channel surfers," the
value of such a system is clear.

IL DEFTS0009133

-.::t

C")
T"""

(J)

o
o
o

Figure 1: Hot Now Application Client/Server Diagram

Cf)

l-

1.
Alert Sender

LL
W

o

2.
Alert Receiver 1
Sensitivity Filtering

Alert Message:
[URL, senderID, Comment]

2.
Alert Receiver 2
Alert sent to clients within inferred interest groups:
[I jst of Inferred Interest Groups, URL, Comment]

Sensitivity Filtering

2.
Alert Receiver 3
Sensitivity Filtering

Internet

Alert Message

Filtered Alert Message

Queries:
-Sender Interest Groups
-URL Interest Groups
-Receiver Interest Groups

Hot Now Server
Inferred Interest Group
Filtering

Database

Figure l 2. Hot Now Application Server Side FlovJ Chart - BitSurge Protection
Legend
vs = Number of Viewers of a
Web Video x
conn = Number of connections
allowed by Server hosting Web
Video X

o

Start

Request to Server:
Serve Web Video X

Is it first Request to
Serve Web Video X

No

Yes

Cache Web Video X

Get Number of Viewers (vs)
for Web Video X

Is

vs > conn

No

Serve Web Video X

,..-_--L..._-,

Yes

,---.1....--...,

Serve Cashed Web
Video X

IL DEFTS0009135