Future TV service
Here are my thoughts how i see it:

Current cable services consist from big number of channels, this number
so big (300-500) that it is really hard to overview them in reasonable
time and remember what is where going. DVR/PVR devices appeared to
overcome that, so people can search tvguide daata and schedule to record
the programs they would like to watch later.

So to go further, let's assume that new video service wil l be 99% on
demand, just 1-5 channels, leaving other channels for live info like
weather/news or events. On 1 video channel, customer would browse the
catalog, use search just like google to find anything that matches and
see description/meta info about video content. Just like current Video
On Demand but not just for movies, for everything. All the shows,
documentaries and other programs are pre-recorded and broadcasted via
sattelites to the video providers. The same content to everybody, movies
are duplicated by numerious channels, except very specific channels like
Food, Sport, History the rest of general channels are showsing the same
content, just different TV guide grid/schedules.

So, service provider needs to be connected to aggregator like TVN,
collect all content and save it. It costs money, they charge by volume,
so it may take some time to accumulate enough content to be able to
satisfy almost everybody needs.

On other hand, in this All-On-Demand mode, service provider can charge
by play, every time user clicks on Play button, he is charged some
amount, like 25 sent to 99 cents, like iTunes.

Some content needs to be free of charge, so people can watch it freely
and get accustomed to the system. Actually a lot of content will be
free of charge, everything what is broadcasted now or included into
channel lineup will be free.

Assuming a video provider has 3000 houses, if in every house
Pla y button would be pressed 3 times a day, we can have:

- 3000*25*3 cents = $2250 a day, $67500 a month, $810,000 a year
or
- 3000*99*3 cents = $8910 a day, $267300 a month, $3,207,600 a year

and this is very conservative, some may not be using TV at all, some may
watch more than 3 programs, kids would watch several cartoons a day
easily. For every houshold, not paying montly fee or paying something
like $5-$10 a month plus all on-demand charges may result even in less
amount than current $39-$50 a month for cable services and they can
watch whatever they want whenever they want.

If i watch only 1 program a day i pay maximum 0.99*30 = $29.70 a month which
sounds like better deal, because i come and choose to watch what i like
at this particular moment and it may not be every day i watch TV.

This will require completely different infrastructure and special settop
box(media computer).

For content providers, the revenue can be s plit for every Play press, so
they will have additional income. Once user pressed Play, the system
charges his account and he can watch this program for 24 hours any
number of times. Good deal for kids, they can watch cartoon several
times and parents will not be worried about additional charges if they
allow children watch TV for so long.

Technically speaking as i see it settop box should be small media
computer. Having smart device will help making it more personalized, i t
can keep preferences and some content locally, trace behaviour and
provide customizations, like menus, colors, notifications, people can
connect it to their home PCs and play music or home videos.

Almost everybody nowadays keep some kind of music/video collections, and
if people could collect it, this is what they like and helping them use
it will only make them more happier about this device/service, and allow
them use it as a central media device in the house.

This device will not be DVR/PVR capable because no need to record
programs locally, they are available all the time on the servers. Anyway
content providers are and will not be happy about spreading their
content across people, event if DVR does not allow to distribute it. It
is possible to lower charge every time somebody watch the same movie or
program, until after let's say 3 times it will be free forever or for
the next 6 months or so.

Content should be described, put i nto catalog, made searchable by any
type of meta information, for example if it is a movie, description,
actors, reviews should be stored and available to search. Same for all
kind of programs and mucic files.

On the TV screen, using media settopbox, user can browse catalog, but it
should not have too many genres or be doo deep, it makes it hard after
several levels to keep track and narrow down what i what to find.

Up to 3 levels directory then search or browse the li st or just do the
search by keywords from the top. The key is easy to use navigation and
search capabilities.

Security can be implemented very easyly: SSL for transactions and some
kind of scrambling/ligth encryption for watching. It should be light but
adaptive and mutative. I think about scramble generator, we can generate
hundreds of small libraries which can encode/decode using any types of
enryption schemes. Such binary decoders will be pushed to settop boxes,
so n ext time they press Play, over SSL connection settopbox will receive
Id of decoder scheme and it will load corresponding binary lobrary that
implements it. Every once in a while the system genereates new binaries
with modified or "mutated" algorithms and then updates the settop boxes.
This way, they itself is binary code which lives very short period of
time, so even if somebody cracked one library, it will never be used
anyway. It can be combination of binary code and some kind of key.
Settop box will be closed, no access to it, so the only way to crack it
to capture network traffic which somebody eventually will be doing. So
making it very dynamic will make it harder to crack and in case of break
it will be easy to reload all boxes with new code/keys. This needs more
thought, but currently may sites uses MS DRM code which is SSL plus
certificates over Internet, but in closed network only house owner can
do it and it is easy to track activity if you know who is who.

Storage needs to be huge, for example 1 encoded movie is up to 1.4-2Gb.
to keep 2000-3000 of movies will require 60Tb of storage which is not a
big deal. Several thousand of programs will need to in the catalog as
well, plus every month new serials/soap operas appear, so 10Tb is
a minimum.

The big question is network, 1Gb ethernet on the backbone will not be
able to handle it, 10Gb is the minimum. 1Gb will handle up to 300
simultaneius stream of 3Mb MPE G2. so, 10Gb will handle up to 2000
streams, reserving some bandwith for special needs.

This is straighforward approach, but it can be improved, like, making
the movie to be downloaded if it is small enough, so waiting 10-15
seconds is not a problem, clustering the system, using bittorrent like
protocols to spread the load.