Using Pauls's main arguments (historical perspective) it's not hard to "prove" the opposite: world is constantly moving towards consolidation.

Look: Desktop software development world looked very much like web scene today: one person could invent (and implement) an electronic table or an editor or a basic interpreter or an interesting game and do very well financially (has been done thousands of times). Yet in the early-mid 90s most software got prohibitively expensive to build for a small firm.

Also you can go back to early automotive boom in the US: there were myriads of automotive startups in mid-west (and in Europe too) and look what happened to all of them later. Same can be said regarding telecom, oil and railroad industries.

Developing industry = more startups Mature industry = very few startups

Has always been like that.

* Coase and lower transaction costs. I think this generally favors pg's argument.

* Capital requirements for new companies. With internet startups, these are extremely low. Is this an anomaly that will disappear soon? Is it a new trend that will spread to other fields? Is anything else that has 'gone before' comparable to this field's low capital requirements?

We're already seeing this, of course. Bad laws have real consequences, and it often becomes politically impossible to repeal them. Sarbanes-Oxley has all but shut down the IPO exit. Consider what the expansion of 'know your customer' laws to online commerce might do to startup SaaS businesses, for example, when you consider the huge compliance costs. And there are a million ways to hamstring online startups in the eternal quest to 'protect the children.'

While we've certainly seen a decline in external transaction costs, I think we're also seeing a decline in internal transaction costs. This may eventually spell the demise of the hierarchically rigid organization if organizations become more internally flexible and democratic (this is certainly not yet the case), but as PG points out it won't spell the demise of large organizations per-se.

Capital requirements for software companies might have gone down due to better tools and cheap hardware. However, it's gotten more expensive, and borderline impossible given the regulatory environment, to start any sort of operation that builds, grows, mines, or transports things. I'd bet it's more expensive to start a medical practice or an accountancy now than in the past, too.

And as you say, "Young, fast-developing industries tend to produce rich ecosystems of startups".

These two points imply that the trend is to have many young industries (many more than there are old), and thus many many more startups than big companies.

Further, he mentions how new technologies are game changers. Big companies cannot adapt as fast as smaller ones, and therefore their "death-rate" will be higher.

At the end of the day, I think you're not taking into account the fact that the rate of technological innovation is accelerating. This fact is going to be a large cause of much social and economic disruption (a good thing).

http://steve-yegge.blogspot.com/2007/06/that-old-marshmallow...

In this interview, Steve also points out a point along the lines of the ancestor poster's - that Google encourages employees who are ambitious enough to start "startups within Google".

1. http://blog.stackoverflow.com/2008/10/podcast-25/

Also you can feel the helplessness that management creates perhaps inadvertently in the engineering team.

Large organizations will start to do worse now, though, because for the first time in history they're no longer getting the best people.

I believe this statement to be false. Sampling from my college class and my smart hometown friends, the most common careers are, in order: 1) Law School/Med School 2) Wall St. 3) Consulting 4) Government 5) Academia 6) non-profit 7) other large corporation 8) small business. Perhaps 1% are working for a startup.

This dominance by large organizations is historically weird. For most of American history, the goal of an ambitious person was to own his own business or farm. The really ambitious would aim to turn that small business into an empire. Statistically, the average firm size was far, far smaller in 1900, so most ambitious people would be working for small businesses. Anecdotally, my great-grand parents mostly ran a business or were self-employed. But my father and uncles were mostly "organization men", as are most of my peers.

I think relative to 1960, people today are more likely to work for small organizations. But the "organization man" economy of the 60's was a creation of the New Deal and World War II. We're still trying to unwind the consolidation of that era.

You have (accidentally I realize) selected a very anomalous time and place as your counterexample. In the first half of its history America was practically unique among richer countries in having huge amounts of unsettled agricultural land. So yes, in America between 1650 and 1850, a lot of ambitious people moved west to establish new farms. But broaden the scope to include the rest of the world, and the time to more than that 200 year window, and ask the question: did an ambitious person stay in his local village and farm, or did he leave to seek his fortune?

You can also approach this question from the other direction. Examine a large organization for its time (e.g. the British East India Company, the Medici bank, Cluny, the papal curia) and ask: were the people running it more or less ambitious than peers who'd stayed on the farm?

Either way you get the same answer.

Of course, in places without entrepreneurial capitalism, most ambitious people aimed to work for large existing organizations - usually the government, church, or military. Examples: France before 1815, the papal states, China for its entire history, the late Roman empire, modern Japan, and modern Europe.

It seems part of your thesis is: "For most of history, ambitious people sought to join large organizations such as the church, state, or army. But with the rise of entrepreneurial capitalism, ambitious people instead chose to start or join small, growth oriented businesses"

I agree with this part of your thesis completely. My point of contention is that I don't think entrepreneurial capitalism began in the late twentieth century. Entrepreneurial capitalism began in earnest around the 17th century, and peaked across the Western world in 1914. Then the world wars and the progressive movement almost completely wiped out entrepreneurial capitalism and replaced it with the civil service state/managed capitalism.

The rise of Silicon Valley is a step towards restoration, not a brand new innovation. Nor are venture capital firms anything new. It used to be that banks and angels were the primary sources of venture capital, see the Boston Associates for one example of many: http://en.wikipedia.org/wiki/The_Boston_Associates. But post-New Deal, banks were highly regulated, non-risk taking entities. The 95% marginal tax rates severely limited the supply of wealthy investors. The founders of Silicon Valley had to basically reinvent the wheel and call it a different name.

Establishment elite have used government to pervert the risk-return relationship. You can get relatively low risk high returns as a doctor, for example, because the profession is cartelized.

http://www.2blowhards.com/archives/2007/02/risk_reward_and.h...

Perhaps the closest example I can think of here is the transformation that happened in both Christianity and Islam with the rises of the Catholic church and the Caliphate, respectively. At a certain point their religious function became incidental. They were by and large instruments which brokered in power and influence.

You can see a similar pattern in governments founded through populist revolutions based on ideologies as they've transformed into world powers. Liberty, as such, has for quite a while ceased to be the primary function of the US or French governments.

In the same sense, I believe that the modern trans-national corporation has ceased to be an entity that exists primarily for the propagation of products and services: they are also brokers of power and influence. There was a first colonial phase starting at the middle of the last century where corporations became powerful political influence outside of their own home territories and since the 70s or 80s they've extended to becoming powerful forces even within the machines of the modern super-powers. Drawing another analogy, what I believe the modern renewal of small companies is effectively a reformation akin to what's been seen in religious institutions, where there's a sense of things getting back to the ideals. I feel like it's still too early in history to see if this will be a fundamentally disruptive change, or merely a blip in the growth pattern. Again, history provides some examples there for example in the Wycliffe-ian or Hussite movements, which preceded the more dramatic changes of the reformation and counter-reformation (and before those movements themselves began creeping towards instruments of power). The real reformation may still be yet to come.

Mature industry = chance for disruptive change = developing industry

The whole raison d'être of a large organizational pyramid is cost of processing and transmitting information and handling the associated noise. Middle layers of management filtering and processing information for the upper decision-makers. Keeping the noise down by designing company 'policies', 'strategies' and 'missions' that can be easily communicated to every employee.

A manager who designs the most efficient and scalable organization structure wins over the competition or conquers a new market.

Here comes the internet with zero cost of communication. Here come high level languages that make it cheap to describe the most efficient processes.

Any large organization can in principle be substituted by a code because the organization itself IS a code written in job description language.

Here comes the internet with zero cost of communication - information still costs time (opportunity cost) to produce and consume. It can be designed to be easier to produce/consume, but that costs design resources. The Internet gives zero cost to transmit information.

Here come high level languages that make it cheap to describe the most efficient processes. - this won't happen because people are not interchangeable like Silicon is. You can hire two people that fit the same job description but will fit differently into the organization. They will work differently, interact differently with management and coworkers, produce different results (even if equivalent).

I think the closest we can come to high level business languages are

1) something like design patterns that can't be plugged in directly but can guide implementation of a business process with known trade-offs OR

2) design processes that can be completely specified and automated so human judgement isn't involved

Interestingly, #1 favors talent cultivation (a la Google) and the second leads to outsourcing.

I was in my teens during the latter half of this decade - I'd guess that at least half of the people who read hacker news are too young to have seen them when they came out. If you do watch them now, they'll probably seem as quaint as "Hair".

Don't ask.

http://www.imdb.com/title/tt0096463/

Ironically, Tron was this way also. Flynn sought to run the big tech corporation instead of that arcade. Now, I'd rather be running that arcade than working for the big corporation.

I'm not a historian, but as far as I know from reading various books by Drucker, large corporations didn't exist for much of the last thousand years. In fact, I remember he made the point that looking at the turn of the last century, even the largest business of that day would be considered small to medium by today's standards (and goes on to ascribe that to a lack of management knowledge - it's impossible to manage a huge corporation without management).

If "larger is better" has only been with us for a hundred years, it may well slink back into the shadows sooner than we think.

I think the order in which this will unfold will be highly dependent on what kind of business you're in. Car manufacturers and pharmaceutical companies are unlikely to get small any time soon, at least not until we have perfect simulated models of the human body to get rid of all the human trials, and until we have instant, near-free manufacturing processes available to all. By the time those happen, it seems unlikely that "corporations" will look like they do now....

We are observing the decentralization of economic authority, common throughout history, but usually a side-effect of economic decline and/or societal collapse; maybe more correctly a side-effect of the movement from specialized occupations back to subsistence farming.

This decentralization seems to be different, and appears to be driven by the unprecedented communication, access to information, and wide audience granted by the Internet.

The US didn't experience its dramatic economic growth in the 1800s by assembling a large and disciplined army; instead, it disassembled large and disciplined slave plantations, moving the agricultural center of the country to smaller, more productive farms in the North. (Farm productivity per acre continued to vary inversely with farm size, worldwide, until the 1970s.) I think it was its network of what we would now consider small industrial companies, coupled with its relatively low level of violence (due to geographic isolation), that enabled it to outstrip many of the Great Powers in economic growth.

By contrast, during the Edo period, Japan was an extremely large and disciplined organization --- which produced some remarkable art but also made it militarily and economically weak.

So I think the picture is much more complex than, "The success of a society was proportionate to its ability to assemble large and disciplined organizations." I suspect Drucker and Coase's insights may help to provide a more nuanced view.

Edo Japan is a pretty unique case. A famous historical curiosity, in fact. And indeed the power latent in Japanese society became visible in its unprecedentedly rapid industrialization in the second half of the 19th century.

In your essay you predict that "large organizations will probably never again play the leading role they did up till the last quarter of the twentieth century". Are you suggesting that average company size will revert to a level lower than even 1900? Or are you saying that average company size may remain roughly the same, it's just that economic dynamism will come from startups, rather than corporate laboratories?

When you say that traditionally, bigger organizations were better, do you mean they were better:

a) militarily b) in developing new technologies that make lives better c) in producing goods like trains and cars d) better at generating revenues/profits e) better at attracting the best and the brightest f) producing achievements that are remembered by historians g) all of the above

It is important to clarify this, because some of these measures of success oppose each other. For instance, the Egyptian governments were great at f) but at the expense of b).

I at first took your essay to define success as b) improving technology. But if so, the thesis is obviously wrong. Most of the great inventions of the industrial revolution were created by tinkerers and small enterprises, not large organizations. Just look at the development of the textile technology http://en.wikipedia.org/wiki/Timeline_of_clothing_and_textil... . Almost every single invention from 1600 to 1900 was developed by a startup (and many of these startups were funded by outside investors, Kleiners Perkins has nothing on the Boston Associates http://en.wikipedia.org/wiki/The_Boston_Associates ).

Is there a particular book you read that discusses this idea? I'd be interested in reading it.

It seems that you have a different idea than Coase about what limited the size of companies; he believed that companies stopped getting bigger because as they got bigger, they started becoming less "successful", in the sense of "profitable". What's your alternative explanation?

How are you defining success? How well the company did per-capita (per employee)?

But what Graham means, is corporation in it's true sense of the word as a generic "body". The Catholic Church, Roman and Ottoman Empires, Hanseatic League, various hybrid religous orders such as the Teutonic Order and etc, can all be considered corporations. And in general, larger was better, with occasional shifts and retreats.

I think car manufacturing can be done by much smaller organizations. I suspect that pharmaceutical companies will have to change a lot; for all they spend on research, they simply don't prolong human life enough to justify being 14% of the economy.

If you can't ship your products to large distances, you can't have large manufacturing facilities.

Without all these wonderful communication channels, you can't have your workforce distributed across the globe.

What's interesting is that in the 20th Century the big trend was that people expanded their businesses; now, it seems that people are instead specializing more. Presumably some of the reasons for this are to be found in The Nature of the Firm, but while I don't think Coase has the whole story, I think his viewpoint is helpful here. It's not that people are collaborating in smaller groups than before; we're all still collaborating with everyone else in the economy. It's that the nature of that collaboration is changing from intra-company collaboration to inter-company collaboration. Coase focuses on inter-company collaboration through market mechanisms, but another important kind of collaboration is non-market exchange of information. The vast common body of free software and knowledge that we all share access to is a major reason that we can launch an innovative web site today with one or two people working together for a few months.

If you wanted to launch Reddit (or HN) in 1990, you would have had to write client software, set up a modem bank, and ship starter kits to our clients with a CD-ROM and a modem inside. The fact that you can just buy access to the internet cheaply instead is collaboration through the market. So is being able to buy a multi-gigahertz machine with gigabytes of RAM for US$500.

But in 1990, you would have had to do a lot more. You would have had to drop US$500 on a compiler from Borland or somebody and write your code in C++. (If you preferred Lisp, you could drop US$3000 or so on a MacIvory from Symbolics instead; they weren't dead at the time. Or Lucid Common Lisp on a Sun, maybe; how much did that cost?) Now we can write our software in PLT Scheme or Python instead.

If you wanted to do cool AJAXy effects, even assuming that you could transfer the Web software of 2000 back to 1990, you'd still need to implement all the AJAXy stuff yourself, as I did in 2000, instead of just using Prototype or MochiKit or jQuery. A couple of years earlier, you would have had to use Java.

And today, you can run the whole thing on Linux. In 1990, you would have had to buy SunOS, or HP-UX, or Ultrix, or some other such horrible abortion, and the minimum price for that hardware was around $5000. And you probably would have had to recompile your kernel.

And when you have trouble load-balancing, you can Google to find out how people set up reverse proxies and round-robin DNS and the like, rather than inventing it for the first time and publishing a paper about it.

This growth of the information commons means that you can afford to be much more specialized today. You can focus on the unique aspects of your site, rather than figuring out how to run the first or second global information service.

The pattern I see throughout history is that large organizations endure until some disruptive technology takes them down. (Greek hoplite armies vs. light infantry and archers, armored knights vs. longbows, cavalry corps vs. tanks.)

The modern age has simply accelerated the whole process. Technological change occurs increasingly quickly now, and so these "overturnings" happen more and more often. Because large organizations take time to form, the balance of power has shifted to small entrepreneurial organizations.

alexander the great, napoleon, the japanese (1940ish) don't count because they cannot sustain the expansion

ironically, mongol empire (and its vast expansion) lasted for 3.5 centuries. there must be some other reasons that neither cruelty nor barbarism can explain (in fact, empires based on cruelty/barbarism don't last long)

These three industries all have one thing in common, they pay well and they only require a Liberal Arts degree. My sister graduated from Harvard a few years back and I'd say that every single one of her friends I've met works in one of these three industries. Their conversations of others I haven't met indicate that these three professions are the norm for all their classmates.

Perhaps Stanford is different and perhaps CS majors are as well. But the major trend I see is that the most ambitious people from the most prestigious Universities are going to the places that pay them the most.

maybe a quantitative measure like Z = asset_from_work_now / cost_before_work_then where cost_before_work_then include tuition, rent, etc ... work can be defined as wall street, lawyer, plumber, startup, etc

Z for googlers can be in the thousands to millions, while Z for wall streets might be in the tens or hundreds ... big difference

or course Z can be negative (bankrupt+debt) ... but the huge discrepancy of Z value only reinforces my point, only the brave and ambitious dare to walk the uncertain path; the rest only follow thru the proven, safe path

I think a more abstracted version of the problem would be that, at least in western civilization, products spread much faster than best practices. If Nintendo releases a new videogame system they'll sell millions the first day, and yet 2/3 of American children still don't get their daily RDI of calcium.

The problem is that processes are much more important than products in terms of quality of life, health care, child development, business, education, etc. And while technological progress is increasing exponentially, the rate of best-practice adoption remains flat.

It seems like the American dream is being able to purchase the solution to any one of life’s problems in a big f*ing box at Wal-Mart for less than 200 bucks. Consumers are already really well trained at this, and we need to figure out a way to leverage this behavior to get them to adopt new processes. The problem as I see it is threefold:

1) People don't know about best practices

2) They know about a best practice but they aren't sold on it

3) They're sold on it but it's too hard

So far as I can see it, the only way for America to remain competitive with the rest of the world is to make the adoption of best practices an order of magnitude easier. We might be temporarily ahead of everyone else in terms of getting the smartest kids to go out and join or start their own businesses, but in terms of nearly every other best practice we're falling drastically behind.

The reality is that we live in a society where doctors don't wash their hands before surgery, where shoddy farm practices cause excessive soil erosion, where pregnant women eat fish high in mercury and PCBs, etc. America is certainly getting more high-res in response to change, but is the catalyzing change really technological? Corporations are great at buying stuff, it's the best practices they are really slow at adopting. The fact that corporations are breaking down into smaller units seems to be a synechdoche of society at large falling apart due to the same set of failures.

In it he tries to describe firms and why they exist. His main finding is that the reason that firms exist as large entities instead of just individuals trading labour and services with each other is transaction costs. In an industrial society the transaction costs of many individuals coming together to create and sell a product would be prohibitive. By creating firms the transaction costs are lowered - it is cheaper to create and maintain the structure needed to uphold the firm than it is to locate and purchase services in the marketplace when you need them .

The transaction costs are coming down with the advent of the global information flow - it is much easier for me to locate and purchase the services of an individual coder than it would be to locate and purchase the services of a welder 20 years ago. The transaction costs are especially low in industries that sell immaterial services and goods that can easily be moved around the globe, such as web businesses.

This makes firms increasingly less competitive with individuals, it also explains why some sectors such as the auto and aero industries are dominated by big companies and probably will be in the future. The transaction costs of individuals building a Boeing 747 are just too prohibitive.

Link: http://en.wikipedia.org/wiki/Theory_of_the_firm

PS: When companies that can kill Google, IBM, and Toyota while staying small show up I might agree with you until then I expect things to stay like they are.

(http://www.paulgraham.com/highres.html)

* A theme running through many of pg's essay is that startups are a far more productive use of resources. Pg sometimes uses acquisitions as an example taking the same resources & applying them to a large organisation. The result is reduced productivity. Doesn't that imply that acquisitions shouldn't take place in the first place? A transaction is supposed to take place when it produces some sort of a net surplus.

* Can a startup complex survive without acquisitions?

* If size is a disadvantage, isn't this a disincentive to grow? Wouldn't that put a company in danger of being out competed by smaller players?

* If growth is capped by anti - economy of scale effects, is there still enough incentive for startups?

* Today's startups often float around winner take most areas. Isn't this at odds with small organisations being dominant? The solution to this one seems alarming: Small organisations with huge revenues. If this scales we get a large number of Googles being run entirely by a few hundred employees & responsible for a level of wealth creation grossly disproportionate to their number of employees.

My ideas are pretty uncooked. But what I'm saying is that this prediction might be more robust if it explored a separation startup & big business. IE what does a startup look like in an economy dominated by startups. Foxes can never outnumber rabbits.

IBM, Microsoft, Google--they have one or a few core selling products, the rest of their activities could be explained as self-promotion to those that would otherwise invest their time and skills elsewhere. In short, survival.

Stretching it, such large companies may become countries in the sense that they cover the needs of their citizens. Like a VM on an OS. The lucky ones will have wise benevolent dictators (the best political option in Plato's government evolution stages--not that I am endorsing it.)

I hear they tried that in Somalia. Do you read John Robb?

However, I think there is another trend that reinforces what you say here. Previously there was a strong incentive for risk-averse people to work in large corporations. I'm a risk-averse person and I once worked at a large corporation. With the change in organizational behavior so that layoffs have become a first resort action, instead of an action only appropriate in desperate circumstances, there is little risk-aversion benefit to large organizations. So other than health insurance, large corporations have become a generally inferior class of employers. This is true even if you are not entrepreneurial.

Without writing a thesis on it, I offer the following disconnected observations:

* The US Military, one of the largest human organizations, was successfully attacked at it's headquarters for a total expenditure of less than $400,000 and a few lives, and has not and may never catch the attacker

* The Roman Catholic Church's sex scandals

* To be fair to all religious factions, punch in "baptist financial scandal" into google, or any other large religous denomination

* The disappearence of any sort of financial security in corporate employment or retirement

* The fact that there is no "corporate ladder" that PG refers to, and it has likely existed only in people's hopes and fantasies for two decades or so. The way to advance in most corporations is to have the corporation grow beneath you, or to leave for another corporation and come back to a higher position. Promotions and raises inside one corporation do not meet inflation when averaged over all employees. Loyalty is punished, almost as if in some sense the organization knows it is a bad thing and is trying to kill itself.

* The collapse of joint financial organizations that were once emmensely powerful and efficient: * We used to all put premiums into large insurance corporations, and those insurance corporation used their economy of scale to invest those premiums, such that they paid out something like 109% of premiums as claims and still profited. Today most big insurance corporations pay out less in claims than what they take in via premiums. * We used to all put money into banks, which then made loans and paid interest to the depositors. Now we don't save, and as of a few months ago, banks don't lend.

* The US Federal Government, arguably the largest human corporation, is bankrupt and disfunctional -- it was never a model of efficiency, but in times past it got some things done, built dams and won the wars it fought and etc. Now it does nothing productive at all.

* IBM, Microsoft, GE, GM, Exxon, etc -- the readers here don't need to be told of their failures, but I would additionally point out the trend that they don't want new additions to their organization, and attempt to hire mainly contract, outsourced, non-employee employees these days.

* The United Nations

One thing to note, is that I think if the trend is away from big organizations, the trend of integrating Europe into one big EU is a bad one. The trend of increasing Federalization in the United States is probably also generally the wrong direction.

PS: The Pentagon is just an office building NORAD is far more important.

While the stability of large organizations over time, and the relative importance of all large organizations, are separate concepts, both are dropping. While Mormonism or other religions may grow and shift, in general the trend is toward the less centrally organized sects, and the big centrally organized, heirarchical ones are becoming less so, and becoming more run by the lower levels of the organizational pyramid.

Of course the Pentagon is not just an office building. Any reasonable person would believe that one of the hijacked airplanes could have been piloted into a NORAD installation if that had suited Al Qeada's purposes, but NORAD is only useful against other dying behemoths of giant organizations, such as the soviets, and thus isn't nearly as important as it used to be, and would never be bothered with by Al Qeada.

I think the reasons that we did not catch or kill Osama bin Laden at Tora Bora are exemplary of the ways that giant organizations fail. We believed we were strong enough to pursue other goals at the same time (Iraq, various shifts in policy under the cover of the Patriot Act, etc) and thus tolerated various internal groups siphoning off resources to persue those goals, and when things did not go as expected, we could not move fast enough to get back on track, and the internal groups had grown too powerful to stop. If we ever catch Osama it will probably be because we pay someone else to do it for us; I think that is loosely analogous to how IBM had to pay Bill Gates to write DOS for them, because their own internal politics and bureaucracy made it impossible for them to themselves.

One bullet point I left out of my list above, is higher educational institutions. I remember in the early days of slashdot, the education career questions were all of the type "should I go for a Phd. or settle for a Masters" or "the place I want to attend had Computer Engineering instead of Computer Science, does that matter" and stuff like that. Look at the educational career dicision questions on slashdot or here these days -- it is all variations on "should I drop out" "how hard is it to get a job with no degree" and so on.

It seems you are mistaking your wishes for reality...

It's decentralization, and it's been occuring throughout history. For example, equality is a form of decentralization that has taken quite a while to manifest. You've also got decentralization in information, labor (specialization), and housing.

The next one that we desperately need is energy decentralization: we need to have smaller, agile, and more incremental methods of generating energy. The tech is coming along so we'll be able to do this, and some of the tech will only work well if it's not scalable.

My $.02

- The Internet is different from other fields because of low barriers to entry and the ease of collaboration between groups (low transaction costs). This means that even if our field grows very large (and I think it will), the principles underlying its success might not successfully spread to other industries.

- Another thing that might happen is that large organizations gain enough power that they can control the Internet. If they're able to do so and prevent the rise of alternative, more free networks, then hegemony could come to dominate this industry too, as it did all the others.

So I would agree insofar as to say that startups will become the de facto standard method of commercializing new technology, but I don't think they will ever represent a significant portion of the economy as far as the number of people employed. Having such a small percentage of the population involved in startups doesn't afford them much opportunity to affect social norms on a large scale.

Short of complete automation, big organizations will always be best at doing commodity things like picking up your trash and making your shirts. Even once those particular tasks are automated, big organizations will grow (often from startups) elsewhere to tackle new commodity goods and services. Once something is no longer cutting edge, it'll be provided by a big organization.

Startups will always disrupt the way big organizations currently do things, but by being successful they will just become big organizations themselves.

With the advent of open hardware and 3D printers, I think you can take out the word Internet and have the statement still be true.

In my opinion there is not such a huge cultural gap that's responsible for the different startup rates.

Few (if any) cultures are so deeply authoritative as to be anti-success.

The valley's lead is much more the result of rather simpler causes. People - a whole lot of smart ones, two recession proof large smart people producers (the universities), and funding lots and lots of funding.

That's where the lead comes from and it will be difficult to close it. Culture doesn't have anything to do with it.

I would argue that most cultures, even in California, are too authoritarian and inflexible to really cultivate productive, disruptive change. For all the talk about innovation, it is genuinely difficult to argue against the status quo. It is very, very hard to change one's frame of mind, and people tend elsewhere to defer to expertise and experience over ideas with are promising, make sense, but are not so fleshed out. In the Bay Area you will at least find an audience, if not easily convince them. Being anti-authority is obviously insufficient for startup success, though.

A good quote to keep in mind:

"...I would design my own, fresh, without knowing how other people do it. That was another thing that made me very good. All the best things that I did at Apple came from (a) not having money and (b) not having done it before, ever. Every single thing that we came out with that was really great, I'd never once done that thing in my life." -- Steve Wozniak

We are currently encountering much existing knowledge in engine design. Some of it is undoubtedly wise, but a lot of it is probably out of date. We have the chance to experiment. But it would be genuinely impossible to do such a thing in most auto companies today.

summary: http://en.wikipedia.org/wiki/The_Nature_of_the_Firm

actual essay: http://web.cenet.org.cn/upfile/30998.pdf

"An ambitious kid graduating from college now doesn't want to work for a big company. They want to work for the hot startup that's rapidly growing into one."

Thats a generalization and I believe its untrue and biased towards talented, ambitious young people Paul meets. Sorry if I'm wrong, but if that was to convince me, I'd need a proof. From my point of view, the rest of the world is still at sort of peak of belief in the "corporate world" (but I dont have any research doc to reference as well :)).

"Fifty years later, startups are ubiquitous in Silicon Valley and common in a handful of other US cities, but they're still an anomaly in most of the world."

Well thats what I dont understand the most. If the whole article talks about the huge positive impact that startups are going to have on economy, let me ask Paul: what has changed since then ("fifty years [ago]") that makes you believe that NOW is the moment startups are going to change the whole world so much? What was the reason you wrote (published) the article NOW?

Another point to bear in mind: more small companies means more outsourcing, particularly for legal, HR, IT support, and so on. We should also see masses more companies using shared business premises, with shared receptions and so on.

Small companies should also mean more teleworking, since (1) people in small companies are generally more open to the idea of it, and (2) it's easier to keep track of what everyone in a smaller company is doing, even if they are working remotely, and (3) it means small companies can have international representation which may be one guy in a specific Asian country working from a bedroom-office.

"... your strategy is to be as hyperconservative and hyperaggressive as you can be instead of being mildly aggressive or conservative. Instead of putting your money in 'medium risk' investments, you need to put a portion, say 85 to 90 percent, in extremely safe instruments, like Treasury bills—as safe a class of instruments as you can manage to find on this planet. The remaining 10 to 15 percent you put in extremely speculative bets, as leveraged as possible (like options), preferably venture capital-style portfolios."

It seems that going to work for a startup is the exact opposite of this.

I.e., it would be better to be an investor in many startups rather than working solely on/for one startup.

My understanding is that to benefit from "economies of scale", you don't necessarily have to be big in size. You can be a two people startup serving a huge user base with a small margin. It's possible to make huge profits in this scenario because of economies of scale in terms of the size of the market served.

Your use of the terms seems to indicate that you can use economies of scale to your advantage only if you are big in size:

Those who bet on economies of scale generally won, which meant the largest organizations were the most successful ones.

But in the late twentieth century something changed. It turned out that economies of scale were not the only force at work. Particularly in technology, the increase in speed one could get from smaller groups started to trump the advantages of size.

It could be that PG meant that large corporations were more willing to take these bets as they had less to lose than a small company. Taking advantage of economies of scale will generally require some investment upfront as well as adding something to your fixed costs (building a new plant for example). If things don't pan out, you're left with the new expeneses, debt, etc. The financing gap between big companies and small companies has shrunk a lot of the years. During the times that PG is talking about, a small company would have difficult time getting the same amount of financing as a large company in a comparable amount of time. I doubt a small company back then could scale at 10% of the speed that a small company could now.

Btw, Wikipedia seems to be more or less in agreement with my stand here: [http://en.wikipedia.org/wiki/Economies_of_scale] "Economies of scale are the cost advantages that a firm obtains due to expansion. Diseconomies of scale are the opposite. Economies of scale may be utilized by any size firm expanding its scale of operation."

OK! It seems I missed the cost advantage part; i.e., the term "economies of scale" is used only if there is a cost advantage to be gained from expansion. Is that right?

Economies of scale usually refers to how the cost of producing a unit decreases as you are producing more units at a time.

Taleb calls an endeavor scalable if your reward can increase practically unboundedly with little no extra effort. Examples include pop stars and websites.

In any case, it's probably easier to find loopholes, like startups, or being an early arrival to an industry, or starting at a smaller company that inflates later, or tunneling your way between points in the corporate topology, by switching from one company to another, if you're ambitious. But betting against statistics never pays.

A startup can have a goal besides becoming a big company or being acquired by one. If the company meets that goal, it is successful.

the mongol empire was the game changer [adopting siege machine, biological weapon (throwing infected corpse thru enemy's wall -- the seed of black death), religion (islam, christian, budhism), trade (silk route, 'pony' express), and the rise of creative class]

it conquered the world with bow and arrow, lasted 3.5 centuries (longer than the nuclear powered usa)

barbar/dictator/despotic empire can't/won't last that long -- there's truly something about mongol

This economic trend is directly related to technology and software. One of the problems for big corporations is that more and more is done by machines, i.e., software. However, the singular phenomenon of software is that, the more people working on it, the harder it is to maintain and evolve.

Small groups are much better prepared for success in our present world because they can leverage software. As a single organization grows, so grows the complexity of its internal systems.

PG: Can you expand your thoughts on this? Why do you think this trend is not very far along? What would help in accelerating such a trend?

After startups, this trend could have a significant impact -- meaning folks that start companies/early employees would like to stay longer at such companies, and still be nimble, maybe not to the extent of a startup but not a BigCo. either.

Speculation: This factor may be the difference between life and death for the trend towards high-res. There's today a good chance the US will weaken economically, energy will get catastrophically more expensive, a city somewhere will experience a suitcase nuke.

'should', not 'will'.

A Chinese version of this article is available at:http://www.yeeyan.com/articles/view/pestwave/19240

I suggest retracting this statement. What idea do you have of how the world will look 50 years from now? This may be remembered as your "640kb will always be enough" quote.

.

"The best way to predict the future is to invent it." - Alan Kay

YC is definitely working to prove pg right.

2) what fun is being visionary if you can't say anything bold for fear of being proved wrong?

Also, the 640k quote attributed to Bill Gates is folklore.

http://en.wikiquote.org/wiki/Bill_Gates#Misattributed

http://news.ycombinator.com/item?id=384690

"A few decades" is not a short amount of time. I don't think it's naive to believe that this trend is lasting.

"Do you have anything other than your personal opinion?" I have not stated much...

""A few decades" is not a short amount of time. I don't think it's naive to believe that this trend is lasting."

How do you get from a few decades to an indefinite future? I believe in start-ups, but it doesn't mean that start-ups are the ultimate sort of company forever. Maybe, and this is just an example, maybe in twenty years from now, corporations will finally learn how to identify smart people, and give them what they need, and to quantify their performance, and all the other things that they're lacking, and start-ups will not have many advantages on them.

In summary, even though start-ups are great now, we shouldn't become fixated on start-ups.

In this stage we finally get responses to what was said, rather than how or by whom. The lowest form of response to an argument is simply to state the opposing case, with little or no supporting evidence.

This is often combined with DH2 statements, as in:

I can't believe the author dismisses intelligent design in such a cavalier fashion. Intelligent design is a legitimate scientific theory.

Contradiction can sometimes have some weight. Sometimes merely seeing the opposing case stated explicitly is enough to see that it's right. But usually evidence will help."

http://www.paulgraham.com/disagree.html

In retailing, Wal-Mart is crushing their competitors, who are also extremely large but they do not have the economies of scale that Wal-Mart. (In addition, Wal-Mart has better management.)

In some cases, large companies succeed because consumers want the perceived value of purchasing from a large company. Dell and HP basically own the server and desktop computer market, and there is no reason to assume this will change in the next ten years. People feel that Dell will not go out of business.

One of the industries hurting most right now is the automobile industry. GM and Chrysler say they will run out of money unless the U.S. government provides several billions of dollars in aid soon. Ford says it does not need government financing but it has pledged all of its assets in a collateral-based loan. Yet almost no one is willing to buy a car from a small automobile company, they are viewed as risker than GM or Chrysler.

In choosing software companies, one factor sophisticated purchasers look at is, "How likely is this company to be in business 5 and 10 years from now?" When you commit to a software package, you are making a huge investment in time and energy, far beyond the purchase price of the software. You want to know that in the future, the software company will continue to enhance the product and will be around to provide technical support. This is why companies, for example, purchase software from Oracle, even though Oracle is ridiculously. People expect Oracle to be in business ten years from now. Yes, open source to some extent mitigates this risk, but putting aside very successful open source projects (Linux, Apache, MySQL, how do I know that they will be around ten years from now. Do I really want to writing the code myself?)

Law firms have consolidated tremendously in the past decade. Large clients want one firm that can provide experise in dozen of areas of the law and can handle complex transactions that span the globe. In Boston, ten years a 100 lawyer firm was considered to be a large sophisticated firm. Now such firms (at least on the corporate side) have merged with other firms to create 500 attorney firms.

In public accounting, the Big Eight has consolidated into the Big Four. Clients want the brand name that only a Big Four firm can provide.

There is no clear trend here. In some industries, consolidation is taking place and will never stop. In other industries, it is the opposite effect.

James Mitchell www.bostonconvivium.com [email protected]

I think that the early 1900's were a more economical natural time. There were some big corporations, but the economy was largely dominated by small businesses and startups. The 1960's were the anomaly, a direct result of the creation of the American mega-state.

Today the government and large government connected corporations have calcified. Look at NASA. It's not sexy to work there anymore because everyone knows that it hasn't done a single interesting thing since it put a man on the moon. But because it is a government program, it cannot fail and be replaced with something more dynamic. It's a zombie organization, not really alive, but unable to die.

But till recently this was an anomalous route that tended to be followed only by outsiders. It was no coincidence that the great industrialists of the nineteenth century had so little formal education. As huge as their companies eventually became, they were all essentially mechanics and shopkeepers at first.

In 1900, getting an education was anomalous. Most smart people were mechanics and shopkeepers, and if they were ambitious and talented, they created successful startups. The hacker in 1900 viewed a college degree like a hacker today views an MBA - as an expensive way to rot your brain. And finally, for those who got engineering degrees, I doubt starting a company was that uncommon. Off the top of my head, the very first student at Stanford ( Herbert Hoover) did indeed start his own company. So the tradition has been there for a long time.

Now I would guess that practically every Stanford or Berkeley undergrad who knows how to program has at least considered the idea of starting a startup. East Coast universities are not far behind, and British universities only a little behind them.

In the 2000's your average ambitious Ivy league graduate wanted to work for Wall St. The rest would go into consulting, academia, or get a profession degree. Very few people think of doing startups. If the Wall St. bubble never comes back, and graduates are forced to make an honest living, perhaps there will be a trend towards doing more startups.

Fifty years later, startups are ubiquitous in Silicon Valley and common in a handful of other US cities, but they're still an anomaly in most of the world.

Startups are an anomaly in Europe, Japan, and Korea because of government regulations. China is filled with startups because it has the freest market in the world. Startups are common in Sillicon valley because it is the home of the tech industry, which is the most unregulated industry in the United States.

Do you mean "arbitrage" or do you mean "triage"?

http://h3h.net/2008/12/economy-of-software-maintenance/

Even if Internet-related applications only become a tenth of the world's economy

A tenth? Internet applications? A tenth of the world's economy could be Internet applications? I'm a bit flabbergasted by the thought. I mean, I guess...if entertainment spending comes to be merged into the Internet applications category, and media, and toys and games, and travel, and education, etc. I guess the argument can be made that a tenth of the investment an average business makes (regardless of what the business actually does) in the future will be on Internet-related technology.

But, I think we're still a long way from that, aren't we? And it doesn't alter the basic realities of the economy, and where people spend most of their money.

How much do grocery stores and food producers spend on Internet applications? Certainly not 10%. Real estate and other big ticket item sellers like cars and boats? Is it 10%? Seems very unlikely, though maybe with advertising becoming more Internet focused one could say that their advertising budget will eventually be going to the Internet. Food and houses and other necessities are where 1/3 to 2/3 of the average person spends their money, worldwide. So, what else is there? Energy. I suspect energy companies spend maybe a tenth of one percent of their annual budget on Internet-related expenses, if that much (though they spend a lot on technology, in general).

Unless we imagine that the percentage of money people spend on these things will drop significantly in the near future relative to entertainment, toys, sports, and knowledge-related activities (and all of those things become dramatically more Internet focused--which is the believable aspect of this scenario), a 10% Internet application economy future is hard for me to visualize.

Of course pg is theorizing that the wealth being created is independent of existing systems of wealth generation. Being a libertarian, I'm sympathetic to that notion. But, there needs to be a productivity boost created by new technologies for it to literally create wealth. It needs to save people time or make some sort of production or transit more efficient. Entertainment doesn't produce wealth, IMHO, it merely redirects it...and some percentage of modern Internet applications are entertainment. I'm not saying that's a bad thing...just that I believe when the wealth being "created" isn't actually in response to production or increased efficiency, I suspect it's squeezing a balloon rather than blowing more air into it.

And, the natural fallout of that is that if the brightest young lads and lasses are going into Internet applications, instead of taking part in the physical world, we might end up slowing the growth of wealth generation in the world rather than accelerating it. Particularly if they're mostly doing pure entertainment applications.

Luckily, as long as we have a free market, it will correct for those kinds of mistakes, even if it takes a while for us to recognize the corrections (and even if the mistakes get subsidized across several years because investors are slow to react to shifts in the landscape).

look at wikipedia , skype , and linux. another example:worldwide yearly vaccine market is $11 billion , worldwide daily oil market is $6.5 billion.

the problem with those going to internet apps is not that they chose internet apps , because internet and software apps hold enormous potential value. the problem is the things they chose to implement , are not really valuable. but that's maybe a basic issue related to the consumer culture.