A startup's initial business plan doesn't matter that much, because it is very hard to determine up front exactly what combination of product and market will result in success.

By definition you will be doing something new, in a world that is a very uncertain place.  You are simply probably not going to know whether your initial idea will work as a product and a business, or not.  And you will probably have to rapidly evolve your plan -- possibly every aspect of it -- as you go.

(The military has a saying that expresses the same concept -- "No battle plan ever survives contact with the enemy."  In this case, your enemy is the world at large.)

It is therefore much more important for a startup to aggressively seek out a big market, and product/market fit within that market, once the startup is up and running, than it is to try to plan out what you are going to do in great detail ahead of time.

The history of successful startups is quite clear on this topic. 

Normally I would simply point to Microsoft, which started as a programming tools company before IBM all but forced Bill Gates to go into the operating system business, or Oracle, which was a consultancy for the CIA before Larry Ellison decided to productize the relational database, or Intel, which was a much smaller company focused on the memory chip market until the Japanese onslaught of the mid-80's forced Andy Grove to switch focus to CPUs.

However, I've recently been reading Randall Stross's marvelous book about Thomas Edison, The Wizard of Menlo Park.

Edison's first commercially viable breakthrough invention was the phonograph -- the forerunner to what you kids know as the record player, the turntable, the Walkman, the CD player, and the IPod.  Edison went on, of course, to become one of the greatest inventors and innovators of all time.

As our story begins, Edison, an unknown inventor running his own startup, is focused on developing better hardware for telegraph operators.  He is particularly focused on equipment for telegraph operators to be able to send voice messages over telegraph lines.

Cue the book:

The day after Edison had noted the idea for recording voice messages received by a telegraphy office, he came up with a variation.  That evening, on 18 July 1877, when [his lab's] midnight dinner had been consumed... [Edison] turned around to face [his assistant Charles Batchelor] and casually remarked, "Batch, if we had a point on this, we could make a record of some material which we could afterwards pull under the point, and it would give us the speech back."

As soon as Edison had pointed it out, it seemed so obvious that they did not pause to appreciate... the suggestion.  Everyone jumped up to rig a test... within an hour, they had the gizmo set up on the table... Edison sat down, leaned into the mouthpiece... [and] delivered the stock phrase the lab used to test telephone diaphragms: "Mary had a little lamb."

...Batchelor reinserted the beginning of the [strip on which the phrase had been recorded]... out came "ary ad ell am."  "It was not fine talking," Batchelor recalled, "but the shape of it was there."  The men celebrated with a whoop, shook hands with one another, and worked on.  By breakfast the following morning, they had succeeded in getting clear articulation from waxed paper, the first recording medium -- in the first midnight recording session.

...The discovery was treated suprisingly casually in the lab's notebooks...

It was a singular moment in the modern history of invention, but, in the years that would follow, Edison would never tell the story the way it actually unfolded that summer, always moving the events from July 1877 to December.  We may guess the reason why: in July, he and his assistants failed to appreciate what they had discovered.  At the time, they were working feverishly to develop a set of working telephones to show their best prospect... Western Union... There was no time to pause and reflect on the incidental invention of what was the first working model of the phonograph...

The invention continued to be labeled in the notebooks with the broader rubric "speaking telegraph", reflecting the assumption that it would be put to use in the telegraph office, recording messages.  An unidentified staff member draw up a list of possible names for the machine, which included: tel-autograph, tel-autophone, "chronophone = time-announcer = speaking clock", "didaskophone = teaching speaker = portable teacher", "glottophone = language sounder or speaker", "climatophone = weather announcer", "klangophone = bird cry sounder", "hulagmophone = barking sounder"...

...In October 1877, [Edison] wrote his father that he was "at present very hard up for cash," but if his "speaking telegraph" was successful, he would receive an advance on royalties.  The commercial potential of his still-unnamed recording apparatus remained out of sight...

[A description of the phonograph in Scientific American in early November] set off a frenzy in America and Europe.  The New York Sun was fascinated by the metaphysical implications of an invention that could play "echoes from dead voices".  The New York Times predicted [in an eerie foreshadowing of their bizarre coverage of the Internet in the mid-1990's] that a large business would develop in "bottled sermons", and wealthy connoisseurs would take pride in keeping "a well-stocked oratorical cellar."

...Such was the authority of Scientific American's imprimatur that all of this extraordinary attention was lavished not on the first working phonograph made for public inspection, but merely a description supplied by Edison's assistant.

...By late November, Edison and his staff had caught onto the phonograph's commercial potential as a gadget for entertainment... a list of possible uses for the phonograph was noted [by Edison and his staff], assembled apparently by free association: speaking toys (dogs, reptiles, humans), whistling toy train engines, music boxes, clocks and watches that announced the time.  There was even an inkling of the future importance of personal music collections, here described as the machine for the whole family to enjoy, equipped with a thousand [music recordings], "giving endless amusement."

...The first actual model, however, remained to be built... On 4 December 1877, Batchelor's diary laconically noted, "[staff member John Kruesi] made phonograph today"; it received no more notice than the other entry, "working on speaking tel", the invention [for telegraph operators] that continued to be at the top of the laboratory's research agenda...

...On 7 December 1877, [Edison] walked into the New York office of Scientific American, placed a small machine on the editor's desk, and with about a dozen people gathered around, turned the crank.  "How do you do?" asked the machine, introducing itself crisply.  "How do you like the phonograph?"  It itself was feeling quite well, it assured its listeners, and then cordially bid everyone a good night...

...Having long worked within the world of telegraphic equipment, [Edison] had been perfectly placed to receive the technical inspiration for the phonograph.  But that same world, oriented to a handful of giant industrial customers, had nothing in common with the consumer marketplace...

The story goes on and on -- and you should read the book; it's all like this.

The point is this:

If Thomas Edison didn't know what he had when he invented the phonograph while he thought he was trying to create better industrial equipment for telegraph operators...

...what are the odds that you -- or any entrepreneur -- is going to have it all figured out up front?

Summary:

A startup's initial business plan doesn't matter that much, because it is very hard to determine up front exactly what combination of product and market will result in success.

Don't assume. Keep learning as you go.

About the archive:

http://pandawhale.com/post/17891/about-the-pmarca-archive-from-marc-andreessens-blog

pmarca Guide to Startups, Part 1 -- Why Not To Do a Startup:

http://pandawhale.com/post/18395/the-pmarca-guide-to-startups-part-1-why-not-to-do-a-startup-pmarca-archive

pmarca Archive on PandaWhale: 

http://pandawhale.com/ifindkarma/pmarca-archive