We all know about the groundbreaking work from Chris Anderson from Wired about the Long Tail. In theory it makes a ton of sense – on the web, companies have no inventory costs and can stock as many titles or products as possible and that over time the one-offs or misses can generate as much or more sales than the hits. As you can imagine, this Long Tail meme gets mentioned by many an entrepreneur that I meet and saying "we are going after the long tail of X" is almost as popular as saying "I’m a Web 2.0 company." I have not read the book or the data, but as I said, in theory it sounds great. You could even extend this long tail concept to user generated content. For example, YouTube could be like the long tail of video – people get to see new content which would never sell at any traditional bricks and mortars store and YouTube has the opportunity to make money off all of this Long Tail content.
As for the Long Tail, the only question one can ask is when will it happen vs. if it makes sense or not. In today’s Wall Street Journal, Lee Gomes (see his article here – sorry, requires login) challenges the timing of the Long Tail and comes up with some interesting data.
"By Mr. Anderson’s calculation, 25% of Amazon’s sales are from it’s tail, as they involve books you can’t find at a traditional retailer. But using another analysis of those numbers – an analysis that Mr. Anderson argues isn’t meaningful – you can show that 2.7% of Amazon’s titles produce a whopping 75% of its revenues. Not quite as impressive.
Another theme of the book is that "hits are starting to rule less." But when I looked online, I was surprised to see what seemed like the opposite. Ecast says 10% of its songs account for roughly 90% of its streams; monthly data from Rhapsody showed the top 10% songs getting 86% of streams."
Lee has a few other examples and one of the most interesting ones is when he states that when Chris looked at the data 2 years ago for eCast that 2% of songs did not play every quarter and now with a much larger inventory that number has risen to 12%. Maybe eCast just had the hits up in the first place? In short, Lee Gomes concludes that the Long Tail may be true but it will also take a long time before it happens.
From my perspective, I do believe we still live in a hits driven world but that it is definitely changing. In addition, if you apply the concept of the Long Tail more broadly to concepts ike YouTube, etc. then it is happening today. Regardless of what you think, Lee’s article is one of the few that I have seen challenging the Long Tail meme that we all want to believe.
UPDATE: Since I posted from the train this morning and have been in meetings for most of the day, I did not get to see Chris Anderson’s thoughtful response to Lee Gomes. Here is an excerpt from Chris’ post:
What it does say is that the current data at Rhapsody, Netflix and Amazon show that the tail amounts to between 21% and 40% of the market, with the head accounting for the rest. Although I don’t discuss this in detail in the book, in the case of Rhapsody, the trend data suggests that the tail (as defined above) actually will equal the head within five years. Which is why the language Gomes cites from the book jacket is actually all phrased in the future conditional tense ("What happens when the combined value of all the millions of items that may sell only a few copies equals or exceeds the value of a few items that sell millions each?"). I asked him to quote the jacket copy in full context, but it apparently wasn’t convenient to his thesis to do so, so he didn’t.
From this post, it seems that Lee misquoted Chris and that Chris agrees that it will take some time for the Long Tail to outsell the hits.
UPDATE 2 – Please read Lee Gomes’ comment to my blog post below where he clarifies his thinking on the article and stands firm on his position especially in relation to Chris Anderson’s rebuttal and my commentary where I suggest that he may have misquoted Chris about the impact of the Long Tail.
Ed, I usually don’t respond to blogs, not because I don’t value them enormously — I do — but simply because I write for a pretty big outlet myself, and think that once I have my say about something, I should shut up and let others have theirs. I need to comment, though, on your suggestion that I might have misquoted Chris. As I hope you appreciate, that is one of the worst things a journalist can do, even (or especially) when writing about a person whose views are being subject to scrutiny.
Here is how I described the book’s premise about this matter: "In the book’s main sections, Mr. Anderson writes that as things move online, sales of misses will increase — so much so that they can equal or exceed the sales of hits." Note that it is written in the future conditional tense, exactly like Chris says own his sentence is. I never said that Chris said that misses were currently outselling hits; my point was simply that considering all the to-do he makes about this in the book, I was a little surprised that he didn’t have any current examples. Had I had more space than I do for my column, which recall runs in the print paper and thus is limited to around 850 words, I would have happily quoted Chris’ entire sentence, as well as this other one from the jacket. "Using the worlds of movies, books, and music, he showed how the Internet has made possible a new world in which the combined value of modest sellers and quirky titles equals the sales of top hits."
As for the suggestion, not yours, that I misunderstood Chris’ methodology: I know perfectly well how he made the calculations he did, and explained them (I hope) very clearly in my piece. I added, though, that there was another way of looking at those same number, making it clear to my readers that Chris did not think that second method was meaningful. At least I gave readers a choice between two methods; the book didn’t even acknowledge that some other method existed.
Thanks for letting me have my say, Ed.
Lee Gomes
Wall Street Journal
Ok-this is done and I thank Lee for questioning the Long Tail meme and stirring the pot as I believe this healthy debate will only improve our thinking and analysis around this concept. Of course, I am curious to see the how the data around the Long Tail evolves as time passes as this transparency will help all of us get a better understanding of the timing and true impact of the tail in certain markets.
Ed, you might want to read Chris’ response to the Gomes piece. He clarifies certain points which are noteworthy. I’m not trying to support Chris, but if someone is going to come after his work, they should do so w/a better understanding of the underlying statistics and perspectives that Chris has laid out.
Here’s the link to Chris’ response:
http://www.longtail.com/the_long_tail/2006/07/factchecking_my.html
Ed, I usually don’t respond to blogs, not because I don’t value them enormously — I do — but simply because I write for a pretty big outlet myself, and think that once I have my say about something, I should shut up and let others have theirs. I need to comment, though, on your suggestion that I might have misquoted Chris. As I hope you appreciate, that is one of the worst things a journalist can do, even (or especially) when writing about a person whose views are being subject to scrutiny.
Here is how I described the book’s premise about this matter: “In the book’s main sections, Mr. Anderson writes that as things move online, sales of misses will increase — so much so that they can equal or exceed the sales of hits.” Note that it is written in the future conditional tense, exactly like Chris says own his sentence is. I never said that Chris said that misses were currently outselling hits; my point was simply that considering all the to-do he makes about this in the book, I was a little surprised that he didn’t have any current examples. Had I had more space than I do for my column, which recall runs in the print paper and thus is limited to around 850 words, I would have happily quoted Chris’ entire sentence, as well as this other one from the jacket. “Using the worlds of movies, books, and music, he showed how the Internet has made possible a new world in which the combined value of modest sellers and quirky titles equals the sales of top hits.”
As for the suggestion, not yours, that I misunderstood Chris’ methodology: I know perfectly well how he made the calculations he did, and explained them (I hope) very clearly in my piece. I added, though, that there was another way of looking at those same number, making it clear to my readers that Chris did not think that second method was meaningful. At least I gave readers a choice between two methods; the book didn’t even acknowledge that some other method existed.
Thanks for letting me have my say, Ed.
Lee Gomes
Wall Street Journal
If a tail grows larger than its head is it still a tail, and is the head still a head?
Can there be two heads? After all two heads are better than one.
That is precisely the conundrum – do you use percentages (Top 10%) or numbers, Top 100. As you know, if you use numbers like the Top 100 and the tail gets bigger then it is very likely that the tail will outsell the Top 100 eventually. If you use percentages and say Top 10%, since the number of titles in the library grows and therefore the number of titles in the Top 10% grows as well, that may very well take a lot longer to achieve than outselling the Top 100.
Ed,
It may be that part of the ‘long tail’ you have wanted to find
is not very visible yet, but a lot of the long tail is visible
today, some of it right in front of you. Yes, cherry makes
gorgeous furniture so that someone wandering in a lush forest
of pine converting to oak and hickory and seeing no cherry may
conclude that there are no trees?
You mentioned Amazon: E.g., I’ve bought books from Amazon,
but nearly all of them have been from way, way out in the long
tail. E.g., I believe that my most recent book from them was
Albert N. Shiryaev, ‘Essentials of Stochastic Finance: Facts,
Models, Theory’, ISBN 981-02-3605-0, World Scientific
Publishing, New Jersey, 2003.
Nice book but not a big seller! For some of why, one
prerequisite is, say,
Ioannis Karatzas and Steven E. Shreve, ‘Brownian Motion and
Stochastic Calculus, Second Edition’, ISBN 0-387-97655-8,
Springer-Verlag, New York, 1994.
for which a prerequisite is, say, the first half of
Walter Rudin, ‘Real and Complex Analysis’, ISBN 07-054232-5,
McGraw-Hill, New York, 1966.
for which good prior reading includes, say, all but the
multi-linear algebra of
Paul R. Halmos, ‘Finite-Dimensional Vector Spaces, Second
Edition’, D. Van Nostrand Company, Inc., Princeton, New
Jersey, 1958.
and all but the E. Cartan exterior algebra of
Wendell H. Fleming, ‘Functions of Several Variables’,
Addison-Wesley, Reading, Massachusetts, 1965.
Uh, people interested in the fancy approach to general
relativity should read all of these last two books!
In mathematical finance, possibly of interest in parts of SW
Connecticut and Long Island, in some of the US vernacular,
Shiryaev (a Kolmogorov student!) and also known for
R. S. Lipster and A. N. Shiryayev, ‘Statistics of Random
Processes I: General Theory’, ISBN 0-387-90226-0,
Springer-Verlag, New York, 1977.
R. S. Lipster and A. N. Shiryayev, ‘Statistics of Random
Processes II: Applications’, ISBN 0-387-90236-0,
Springer-Verlag, New York, 1978.
is ‘the man’, but the prerequisites will discourage some
people.
For me, this situation of the long tail is general: My
personal library of books and music (recordings, scores) takes
up most of the storage space in four rooms, and nearly every
item is from way out in the long tail. E.g., when I listen to
my CD of Heifetz playing the Sibelius concerto, I get to
follow along with the score and get to see just how much
Heifetz did that is not explicit in the score! When I play
some of the passages on my violin, what he did gets more
amazing!
‘Pop culture’? Will have a tough time selling me ANY of it.
I had MORE than enough from my time in Memphis. E.g., not
interested in self-mutilation, via body piercing, tattoos, or
otherwise. E.g., from a few random glances of a few seconds
each on TV, it looks like Britney Spears was a pretty teenage
girl, but I have never heard any of her singing. E.g.,
programs have come on prime time TV, been very popular for
years, returned as reruns for years, more than once, without
my ever watching a single episode! Ah, the burden of those
prerequisites! So, for the long tail and the ‘head’, some
people have relatively little interest in the head.
For you, some of the long tail is right in front of you:
There are maybe a few tens of thousands of people in the US
with a serious interest in all three of ‘information
technology’ (IT), starting businesses in IT, and venture
capital. So, for the ‘media’, this interest is definitely out
in the long tail. Since your blog is an example of this
interest, it, too, is in the long tail. You are looking right
at part of the long tail — your blog.
The long tail has many tiny niches. Here is such a niche with
some lessons about the long tail: Some years ago, a friend
and I considered starting a company to make relatively simple
high performance sports cars. I got to be a Full Member of
the SAE (due mostly to my background with airplanes — right,
Memphis) and did some first-cut engineering and business
planning. We got a tour of the company that made the
composite bodies for the original Corvette; the person leading
the tour was one of the main engineers on that work; clearly
they could make nice body parts for us!
For that project, in the end I noticed the need for a lot of
each of engineering and testing, manufacturing space and
tools, supplies and inventory, work with steel, aluminum,
composites, etc., legal efforts, marketing and service
efforts, and beating competition and concluded that the
project was a close call, an okay but not big financial
opportunity. Also stood to go through quite a lot of time,
money, and effort before getting much revenue.
My partner offered to take the first car we built. When I
told him that he was being VERY generous, he was shocked at my
view of the first car! But I had read F. Brooks where he
said, “Plan to throw one away — you will anyhow.” Well, at
times had to wish that Brooks had thrown away IBM’s JCL! My
friend went on to another project, an aluminum yacht nicely
over 50 feet. In the end he was surprised to discover that
Brooks had been correct. Since I already had a background in
IT, I just returned to IT!
Occasionally I did more with that car project: Once I wrote
some software to simulate acceleration by a fairly careful
solution to Newton’s second law F = ma with fairly careful
automotive details. So, given engine torque, gear ratios,
vehicle weight, etc., can calculate acceleration times. Works
nicely for anything that has much chance of being driven on
the street. Once I guessed that could find the stiffness of
space frames by using the implicit function theorem and
multidimensional Newton iteration, so typed the mathematics
into TeX, typed the software into PL/I (terrific for that
problem!), tried it, and saw that it worked nicely! So, can
use this software to calculate the stiffness of automotive
space frames. Maybe I’ll do something similar for suspension
geometry, stiffness of solid objects, and aerodynamic
stability. So far didn’t “throw away” either software effort
or a sports car or a yacht!
Yesterday I happened to see a link to Factory Five Racing.
They are apparently a NICE example of being successful with
what my friend and I had in mind! Maybe I was too
pessimistic? Apparently they have done quite well with a Ford
Cobra and now have moved on to a mid-engine car. Their key
techniques: Use a fairly simple welded tube steel space frame
(I was thinking aluminum, but steel is good enough). For the
interior panels, use just flat sheets of aluminum (I was
thinking sandwiches of balsa, but just sheets are good
enough). For the exterior surfaces, use hand laid composite
just bolted to the frame (what we saw on the tour). Keep the
car light, dry weight about 2100 pounds. Mechanically, keep
the car dirt simple (start by thinking high end go kart but
with a V8 engine, a body, and street legal lights, brakes,
etc.) which is a BIG help in keeping the car light, fast, and
cheap. In that narrow market niche, fast and cheap sell and
also trump concerns about simple. Use a Detroit V8 engine
with high but not wildly high performance. Can do well with
just a cast iron small block; don’t have to use an all
aluminum block over 500 cubic inches with turbo-charging and
inter-cooling (which can need more fuel than the usual
injectors can supply and which makes transmission selection
more difficult). For nearly all the other parts, use just
common off the shelf (e.g., since the car is so light, don’t
need high end brakes). In particular, for a Cobra, for most
of the parts, get, say, two wrecked late model Ford Mustangs
(one wrecked in the front, one in the back!). The really high
performance comes from the simplicity and the resulting light
weight. E.g., a 2200 pound car with 300 rear wheel horsepower
can have better 0-60 MPH and 1/4 times than famous cars
costing $500,000. For marketing, start with the car magazines
and then have a ‘car club’ where people can meet similar
people! From two hours on the Internet yesterday, some nice
lessons!
There are two examples of the long tail here: First, I just
stumbled onto Factor Five Racing. I should have known about
them a long time ago. They are out in the long tail of the US
economy. Second, they have some nice information on their
engineering — including some nice video clips — which is out
in the long tail of engineering and the media.
Yes, in business, it is crucial to concentrate on revenue.
So, one view of the long tail is that there will be more
revenue in the least popular 90% of the instances than in the
most popular 10%. It seems that this situation has not
happened widely yet and that, as a result, you are concluding
that the business opportunities of the long tail are still in
the future. I beg to differ; let’s continue:
Okay, there is a question here: How would I have known about
Factory Five Racing, not just as a company selling products
but also as a subject in, say, my interests in automotive
engineering and business? How? Hmm ….
For more, in violin, how the heck would I finger some of those
passages in the Sibelius concerto? Is there a video of
Heifetz playing that concerto? When I worked through the Bach
E major Prelude, I did the bariolage passages with fingers 1
and 2 and no actual ‘shifts’ of the hand. When I played for
my teacher, he asked, “How much do I owe you for that
fingering?”. How would others find that fingering?
In this way we can formulate a general question: Given a
person and one of their interests, possibly out in the long
tail of the media, the economy, technology, the arts, etc.,
how can that person find instances they would value highly?
How? Hmm ….
Let’s see what people are doing now: The really good video
clips on YouTube, etc. are not so easy to find. Some of the
clips are so good and so difficult to find that old media is
now commonly running ‘stories’ of the form, “Look at this
great YouTube video clip on ….”. Here old media is guessing
that they have found a clip of broad interest. Okay, but,
especially in the long tail, there is also the serious matter
of more focused personal interests. E.g., for all of the
power of Google, if a person is interested in all of IT,
business start-ups in IT, and venture capital for such
start-ups, then one source of links to instances that might be
valued highly is just the right margin of the home page of
your blog. No doubt you put these links on your home page
because such links are not so easy to find. You are correct;
the links are not so easy to find. People used to do such
things before Google. Now, Google is not doing well letting
people find what they would value highly out in the long tail,
especially in new media, and people are putting up lists of
favorite links again.
Maybe a music school professor has a video clip, in slow
motion, of some of the fingering for some of the more
challenging passages in the Sibelius concerto? A video of
Perlman playing the illustrative passages in Galamian’s book?
Maybe a professor of mechanical engineering has a video of a
great lecture on the finite element method in automotive frame
design or on automotive aerodynamic stability? Maybe an
engineer at a tire company has a terrific video clip, with
information TOUGH to get elsewhere, on sports car suspension
design and tire selection. Maybe a computer science professor
has a video clip of a nice lecture on monotone locking
protocols or ways to do ‘behavioral monitoring’ of computer
servers and digital communications networks with known
adjustable rates of false positives and some good asymptotic
results on rates of false negatives?
Broadly, then, instances of new media out in the long tail
that a person would value highly for one of their interests
are not easy to find.
In this way, we have identified a problem in the long tail;
let’s formulate it: People need a good way to help them find
those instances they would value highly as they pursue their
personal interests — e.g., long tail topics in business,
engineering, the arts. Such a Web site should generate a lot
of traffic. There should be opportunities for focused
advertising. So, maybe the professor who did the video clip
on playing the Sibelius concerto didn’t get much revenue, but
maybe the Web site that let someone find that clip got some
revenue from an ad by someone selling tickets to Tanglewood or
repairing violins.
In all of this, the person’s interests are crucial. There are
some severe challenges here: E.g., in the arts, how to find
instances the person would value highly?
Google has been a revolution in civilization. So, revolutions
really are possible. But much more is needed.
I submit that the long tail is important now, is rising in
importance quickly, and will be of overwhelming importance
quite broadly, that the long tail will liberate us from the
tyranny of the least common denominator, “great wasteland”,
and finding needles we want in huge haystacks and be a
revolution in information and progress in the economy and
civilization. But all along, right on the boundary of what is
feasible, we will have to look carefully to see the
opportunities. We may have to take advantage of the oak or
hickory when what we really had in mind was cherry.
I swear that was the longest blog comment I have ever read. I’m not even sure you addressed the controversy over who said what when. Who said what when is a problem I’m very familiar with.
I’m a believer in the Long Tail as far as me being a Long Tail consumer. What I buy is in the Long Tail and if retailers (online) realize this and cater to me more they have my blessing. I’ve spent years looking for albums or books that were out of print.
As for YouTube it seems to have a lot of T&A or whatever the male equivalent is. I’ve never even thought of searching for engineering lectures or instructional videos for instruments. No one has ever directed me to any videos like that either…
Personally, I believe the long tail will be made up of now-hits, and replace the misses bit by bit. Not a lot of providers are willing to write a blank check for disk arrays, and the space they take up in the data center. After all, if I know that 10,000 ‘misses’ account for 20 square meters in the data center, the necessary power etc, and 15,000 Gigs of disk space, and if I also know that the combined total of these misses account for x in revenue, the calculation is easily made whether to offer these flicks or not. A tail made out of previous hits, will surely generate more long-term revenue. Hence the debate seems too academic. Market forces will turn the tail into a collection of proven, ex-hits. Niche players will pick up on the misses -probably the producers of ‘misses’ themselves who will sell directly to end users.
Ed,
Interesting discussion there. I don’t know if you came across this post from Ajit Jaokar that talks about the ‘long tail’ and revenue models for web 2.0.
http://opengardensblog.futuretext.com/archives/2006/07/web_20_show_me.html
Muskie,
Ed wrote
“As for the Long Tail, the only question one can ask is when
will it happen vs. if it makes sense or not.”
So he is questioning WHEN the long tail will happen. By
“happen” he may mostly mean when some information technology
company selling products in the long tail will be a good
venture capital investment.
You mentioned that YouTube has a lot of junk. Here you are
adding importance to the problem I tried to identify: For a
person to find instances of new media, including videos, that
they would value highly is too difficult. Better means are
needed.
Not everything on YouTube is junk: E.g., as of July 29th,
2006 at
http://news.com.com/2300-1026_3-6095928.html?tag=ne.gall.rbcs
was a CNET article on 10 video clips on YouTube that CNET
found likely of broad interest. Some of the clips were good.
That article also illustrates my point: Of the 10 videos, I
really liked just two: One was a play on some a recent
description of the Internet by Senator Stevens of Alaska. The
other was on some results of computational fluid dynamics and
computer graphics by a professor at Stanford. The first
illustrated again how astoundingly little some of our
national leaders understand about the Internet. I found the
second impressive since earlier in my career I worked with the
Navier-Stokes equations at Carderock. So, I liked two of the
10, 20%. So, again, finding instances of new media that a
person would value highly for one of their interests is not
easy and needs help.
Not all the video on YouTube has to be junk, and not all the
video has to be on YouTube: People can put nearly anything on
YouTube or the similar video hosting services of Google, etc.
Or, people can get other hosting.
E.g., can notice the current WSJ piece:
Moguls of New Media
“The MySpace member with a million ‘friends.’ The receptionist
with a production deal. Some of the Web’s amateur entertainers
are becoming powerful players.”
By JOHN JURGENSEN
July 29, 2006; Page P1
with links to videos at various sites.
Just at YouTube, currently their home page is showing as a
“Featured Video”
Day of the Longtail
“Movie Trailer: In celebration of the publication of Chris
Anderson’s book, ‘The Long Tail,’ The old world of media faces
an invasion from another planet. The horror. The horror. (By
Michael Markman, Peter Hirshberg, Bob Kalsey)”
It’s a riot, and apropos!
There are some good technical videos on the Internet: I
mentioned some I found on sports car engineering via the
Factory Five Racing (FFR) Web site. I’m still trying to
figure out how an FFR Cobra with a cast iron Ford small block
can weigh 2100 pounds while a 2006 Corvette Z06 with a Chevy
small block with aluminum block and heads, aluminum frame, and
some carbon fiber body panels has to weigh 3200 pounds! In
physics, I have some videos I found good by F. Wilczek at MIT
and on S. Coleman’s work at Harvard.
I am sure that there is much more on the Internet than I have
seen and that I would value highly for some of my interests.
E.g., as I mentioned, I only stumbled onto the Factory Five
Racing and their information. For this “more”, too often I
have to stumble onto the instances and need some much better
means.
YouTube, etc., now shows convincingly that millions of
individuals can easily do home movie videos, alas, including
many in bad taste.
So, bad taste or not, we have to conclude that quite broadly
across society people can do videos on nearly any subject in
the long tail: There can be university course lectures,
research seminars, research conference lectures, technical
sales presentations, and much, much more — nearly everything
in the long tail, and that is enormous.
The main point in my post was to respond to Ed’s doubts about
the opportunities in the long tail and say that it is
important in various respects now and that there should be an
opportunity now for an information technology start-up making
money from the long tail — help people find what they want in
the long tail and get paid for narrowly targeted ads.
Again, this thread illustrates that (1) there are some good
instances of new media on the Internet, (2) finding instances
a person would value highly for one of their interests is not
easy, (3) better means of finding such instances are needed.
For ‘the vision thing’, again I believe that instances of new
media distributed via the Internet will help break the media
tyranny of the least common denominator, “the great
wasteland”, etc. and be part of a revolution in civilization.
Not small stuff. Some careful consideration is needed, but Ed
is selling the long tail short too soon.
Vincent,
One of the keys to the rise of new media in the long tail is
the great progress of the last decade toward really cheap disk
space.
You mentioned 15,000 gigabytes (GB) of disk space. Well, can
buy SATA hard disk drives for less than $100 each that have
200 GB of space each and put, say, three, in addition to a
boot drive, in one thin flat 1 RU PC. Assuming 28 such PCs in
a rack and we have
3 * 200 * 28 = 16,800
GB in one such rack. Such a rack is about 2′ x 2′ or 4 square
feet.
You mentioned space of 20 square meters, so let’s convert 4
square feet to square meters:
1 inch = 2.54 cm
1 foot = 12 * 2.54 cm
1 foot = 12 * 2.54 * (1/100) m
1 square foot = (12 * 2.54 * (1/100))**2 square meters
4 square feet = 4 * (12 * 2.54 * (1/100))**2 = 0.372 m**2
So, we have about 1/3rd of a square meter and not 20 square
meters, unless we also include space for electric power, air
conditioning, security guards, bridge, systems management
personnel, executive suites, executive wash room, executive
dining room, CEO’s nap room, executive gourmet kitchen,
quarters for executive gourmet chef and staff, executive wine
cellar, etc., all desirable, I admit, but not really from the
video clips!
You also mentioned 15,000 gigabytes (GB) of disk space for
10,000 instances of video. So, that’s 1.5 GB per video clip.
That’s enough for a 90 minute old movie. So far commonly
video clips are taking only about 10% of that much space.
E.g., I have a MOV file of a clip of a F. Wilczek lecture at
Harvard on S. Coleman’s work in quantum field theory that is
48 minutes long; the file is 124 MB, less than 10% of your 1.5
GB.
For something still more common, the
chad_vader.mp4
at
http://www.channel101.com/shows/show.php?show_id=201
is
14.2 million bytes (MB), just under 1% of your 1.5 GB.
At 15 MB per clip, 10,000 clips would be 150 GB and could be
stored on one hard disk of 200 GB available for less than
$100. The little PC in front of me I built last year has an
IDE drive with 250 GB and two SATA drives with 200 GB each. I
still have plenty of space for 10,000 video clips of 15 MB
each. That PC is in a mini tower case and nowhere near 20
square meters.
For more, Seagate is talking single drives with 750 GB, 750
hours of ordinary video, 125 hours of HD video, and 10 video
streams at once. So, 10,000 HD video clips of 90 minutes each
would be
10000 * 90 * (1/60) * (1/125) = 120
drives. That would be maybe 12 JBOD boxes with 10 disks each
and, still, one or two racks of 4 square feet each and, maybe
1200 video streams at once. For RAID, add one drive per array
and use hot plugging. We’re still a long way from 20 square
meters.