Television is suffering
from a glass ceiling. No, I’m
not talking about opportunities
for women or the depth of
customers’ pockets. I’m talking
about architecture.
The glass ceiling is the architecture
behind the typical
TV interactive programming
guide. Current architecture involves
preloading all needed data
to the set-top box. The data is
then funneled straight to the
user interface. This data feed
allows for a few basic features
in the interface, such as grouping
by channel or sorting alphabetically.
Because the set-top box has
limited memory, there is little
or no scope for introducing
new features that would
require additional data. As a
result, the possible set of features
is highly constricted, especially
in terms of discovery
(e.g., search/browse, similar titles)
and personalized recommendations.
When it comes to developing
and introducing
innovations
into the TV guide, the architecture
poses an obstacle. For
example, given the set-top
box’s memory, it is impossible
to preload five similar titles
and all their attributes for every
title in the catalog. In this
model, there is no viable way
to offer customers one of the
most basic, familiar discovery
options: the ability to find
titles similar to one they’ve
seen and enjoyed. As a result
of the architecture, the TV
guide remains a generic, noninteractive
data feed.
For a way out of this impasse,
TV operators might take
a look at leading Internet-content
providers like Amazon
and Netflix. The architecture of
these services is not based on
preloaded data feeds, of course.
Can you imagine browsing on
Amazon without any server interaction?
Internet services like Amazon
help their customers
choose what to buy by combining
their data with a smart
engine. The engine generates
the data that is appropriate to
the customer, such as recommendations
based on what she
or he is currently looking at or
based on his or her consumption
history or product ratings.
This data is then fed to the user
interface.
In other words, the interactivity
with the server enables a
personalized user experience
in which customers are shown
the data that is most relevant
to them. By contrast, the television’s
data-feed architecture
leaves customers to make
sense of the generic data by
themselves.
Today, the average TV viewer
is faced with hundreds of channels,
thousands of on-demand
titles, and perhaps even more
content on a DVR. The choice
is overwhelming. Moreover,
with on-demand programming
growing faster than any
other type of content, channelsurfing is increasingly obsolete
as a way for customers to find
content they feel like watching
now. Customers need easy-touse
discovery features in order
to enjoy, rather than be intimidated
by, the abundance of
choice. In this context, it is
highly problematic that, due
to their underlying architecture,
current TV guides can offer
only limited discovery and
recommendation features.
Television needs a new
model — you might call
it interactive architecture.
Rather than preloading
a generic set of data, this architecture
would involve constant,
ongoing interactions
with the server, to supply the
data relevant to each user at
any point in time. Bypassing
the limitations of the set-top
box memory, this architecture
would allow for a versatile set
of discovery features and a
high level of personalization.
In this model, both the engine
and the data would be integral
to creating a high quality,
satisfying user experience.
Rethinking the TV guide requires
rethinking the architecture.
The next generation of
TV guides is likely to be usercentric,
presenting each user with
a view of the “universe” of content
options that is uniquely suited
to his/her mood and personal
tastes. And the change in the
TV guide will begin with the
architecture.
Yosi Glick is co-founder and
president of Jinni, a video search
company with offices in Israel
and Portland, Ore.
