Hacker News Reader

Comment by WalterBright | Hacker News Reader

The common presumption with free markets is that companies will grow until they consume the world. But as this article suggests, what actually happens is that companies grow until their internal bureaucracy strangles them.

Google's search business is heavily optimized towards and dependent on selling ads. Competing directly with them in search is not likely to work. But another way of doing search (I don't know what that might be) could completely undermine it.

After all, Google sank Alta Vista and Yahoo with another way of doing search.

View on HN

> [...] until their internal bureaucracy strangles them.

I think that standard microeconomics textbooks state empirical efficiency curves tend to have an inverted U shape, so beyond certain size things get more and more inefficient. Bureaucracy, politics and communication play a significant role here.

This is also what I have observed in academic and industrial labs, which scale far less than companies. Beyond certain size, quite small in fact, most managers become rent seekers and the whole thing collapses in slow motion.

I think that this is a significant problem for society, as it leads to a massive waste of resources. There should be better ways to scale up organizations, or to build federated ones. Biological organisms are much better organized than us. We can do better.

> Biological organisms are much better organized than us. We can do better.

If you’re referring to the fact that an organism like a human body can function with trillions of individual cells, it is indeed marvelous and quite well organized. But, cells and people are different. Cells in your body are all genetically identical, and lack any form of “individuality”. Indeed, they possess genetic fetters that will cause them to commit suicide should they become too unique, and should a cell manage to throw off all of those limitations they will become cancerous and potentially kill the entire host.

Regulating a large and diverse set of individuals is a much harder problem than regulating an army of clones, after all.

I disagree with your assertion that cells in a body are all genetically identical- in addition to any cell-specific mutations (which occur any time cells divide) and epigenetics (which provides every cell with unique modifications), at any time a particular cell's RNA complement is unique and this can be used to cluster cells in a high dimensional space. If you just mean "all cells in a body have chromosomes ultimately descended from the gamete", then yeah, sure, but that doesn't really say anything about the full state of a complex organism (which depends on the instantaneous protein concentrations, damage, etc).

What do you mean by biological organisms? Don’t humans fall under that category?

Saying bureaucracies are inefficient is just stating stating very common knowledge. Yes, large systems are difficult to control. Especially so when we’re talking about creatures like us with emotions, desires, a need to socialize with others, etc.

I’d love to hear what better structure/design you have in mind that would scale better and be less inefficient.

Fundamentally, groups of people need to have a leader in order to cooperate towards achieving some goal that each individual by themselves cannot achieve. And groups of leaders need a higher leader. The optimum group size is up for debate but not this fundamental requirement for this general structure.

How do you make sure you don’t have bad leaders? Sorry you can’t. That’s an inevitable byproduct of being humans and not machines you can precisely measure and score.

My pet theory is that the problems which people often associate with government have nothing to do with the difference between government and private business at all and are simply a product of organisation size. The logic that a small number of decisions makers directing a large amount of labour/capital leads to inefficiencies due to a decision-making bottleneck certainly seems to apply to both (and indeed to wealthy individuals beyond a certain level of wealth).

Strongly agree.

Anyone who's worked in private orgs of comparable sizes can attest to their horrendous inefficiencies and waste. The idea that somehow the private sector is inherently more productive or efficient is simply not bourne out in practice. But it _sounds_ like it ought to be true so the idea won't die. And we're all worse off as a result.

Very true

Also you get what you pay for and governments generally are not paying at the top of the scale. Giant + middle of the road pay is going to get you the same results in a private org!

If you add in the value of benefits, government employment pays very, very well.

They are inherently more productive or efficient because:

1. they have a profit motive, and if they don't show a profit, they die

2. smaller companies continually replace large, bureaucratic organizations

> simply not bourne out in practice

SpaceX can launch rockets at what, 10% of the cost of NASA?

> we're all worse off as a result

How so?

Amazon famously operated (may still, couldn't care less) at a loss.

As someone noted above, both can be true- govt can be incredibly inefficient but so can businesses.

My mother hated working with/for the govt as it made 'firing someone an act of congress'. As if to say neigh impossible.

I'd argue we have all incentives misaligned and that is what causes both govt and private business to become shit-shows.

We have government agencies maxing out their spending every year to keep their budgets growing. We have businesses engaging in moral hazard knowing the government (taxpayers) will bail them out.

We need reform from the top down, inside and out.

> Amazon famously operated (may still, couldn't care less) at a loss.

What's wrong with that? Companies trying to get started and establish themselves operate at a loss, burning their investors' capital. Most companies take many years before eking out a profit.

You said private companies have a profit motive- if not, they die.

I used Amazon as a counter-point. Uber still isn't profitable. So if companies can run for over a decade without making a profit, I'd again say the entire system is broken.

Zombie companies burning capital while driving out possibly profitable competitors is a symptom of rot, not a successful financial system.

Again, my point is that for profit companies are every bit as bad if not worse than government agencies who at least pretend to have oversight.

Once companies reach a certain size, their continued success is driven far more by network effects, brand awareness, anti-competitive moats, and similar factors than they it is by efficiency and innovation.

There are plenty of individual examples in both directions. Instead of SpaceX, compare private insurance companies to medicaire. Medicaire is far more efficient.

The profit motive is part of the problem. It routinely creates perverse incentives.

It has gotten to the point that, rather than seeking to build a great product or provide a great service and make a good profit for doing so, far, far too many companies are seeking to make as much profit as conceivably possible, and making the product or providing the service is simply a necessary evil toward that end. That is part of what has led to the catastrophic financialization of so much of our economy.

Money is not the end. Money is a tool. The end goal should always have been better products and services, a better society, and happier people.

Yes. It seems that capitalism is suffering from a particularly unfortunate case of Goodhart's law [0]:

"When a measure becomes a target, it ceases to be a good measure"

Capitalism is founded upon the idea that the amount the market is willing to pay for something is a good measure of its true value. Which used to be a pretty good measure (and to an extent still is - for now). But deregulation has increased our dependence on this measure, which in turn has lead to an increase in it is being gamed. And over time an increasing share of financially successful businesses are those that exploit failures in the metric rather than those providing genuine value.

[0]: https://en.wikipedia.org/wiki/Goodhart%27s_law

> (and to an extent still is - for now)

Upon reflection, I think that this is also a significant aspect of one major part of the problem today.

In the general case, yes: retail products, for instance, are still subject to market forces such that the price you pay bears some reasonable resemblance to what society considers the value of the product.

But what about ad-supported services? The people consuming the services are not, in fact, paying with money. They have no way of gauging how much the services cost, nor how much they personally are giving up in terms of personal data to indirectly pay that cost.

The personal data sector of the economy is, I believe, a perfect example of a place where Goodhart's Law has run rampant, with the only measures available being "how many users" and "how much annual recurring revenue", and both of these being hopelessly complicated—the former to determine just what is a user, and the latter to determine how the value of the service to its users actually translates into profit.

Not sure how coherent this is, but it seems very relevant.

SpaceX is maybe better explained by The Innovator's Dilemma than by the profit motive.

They have radically different ways of doing launches. These are directly threatening to NASA people wedded to the old ways. In order for NASA to adopt their methods wholesale, they'd have to reorganize completely and probably fire a lot of people. And still do an inferior job, because it would take them years.

Wait 20 years and SpaceX will be just as wedded to the methods of 2022 as NASA's wedded to 1969.

What I said about The Innovator's Dilemma is not the whole story. Governments are inherently more risk-adverse, so they'd prefer to take three times as long if it's perceived as safer.

Right, but we're not comparing the government vs SpaceX, but rather, government vs the private sector as a whole. So if SpaceX stagnates (or Google) then other companies will eventually reach the point where they can outcompete them and things start moving again. But you can't usually outcompete the government regardless of how over-large or stagnant it becomes. That's the fundamental difference.

You can't outcompete the government, but you could outcompete a publicly run business if it was setup as a competitive business with its own separate budget. In that case, the government acts quite like a private businesses investors/shareholders rather than running the business directly. Which seems like a best of both worlds in many cases.

Budget separate from what? The government's own budget? That's not possible, at least not in a western style democracy. If it's truly separate then it's not publicly run in any meaningful sense, and if that's not true then it's just a normal government department - it can't exist separate of government funding so in practice, the government controls it regardless of what the participants may claim.

There's not really any such thing as the best of both worlds IMHO. What exact benefits are delivered by being run by the government? The best arguments are in cases of natural monopolies, but even there, there's benefits to just issuing licenses to sub-contractors (the result is a sort of quasi half private half public business of the type you suggest but nobody claims this model is one to use when not strictly necessary).

I wouldn't put USPS and Amtrak on any list of "best organizations" but they do have a somewhat different structure than, say, the Department of Agriculture.

If I can play devil's advocate for a moment: there are some businesses (passenger rail, universal postal system) that are not attractive to any private entity unless they're heavily subsidized. Nonetheless, there's a public interest in providing them.

As for SpaceX: the tech boom's brought us some people with billions at their disposal, and space exploration IS within reach for them. I don't think there was anyone in the 60s with the money and the inclination to do that.

Some countries do have (theoretically) privatized passenger rail systems. It can be done without subsidies too, the UK nearly managed that for a brief period, but then decided to pay for the construction of huge new railways. If you aren't expanding the network though, subsidy free operation is possible. Of course we can argue they're not really private because in practice the government won't let them fail as a group. But in normal times at least, companies that want to exit the space can give up their license or be outcompeted for them.

SpaceX exists because there's now lots of use cases for going into space. In the 60s indeed going into space wasn't so useful and it happened so early just because of the space race.

Yeah, the problem with passenger rail is, as you said, "the government won't let them fail as a group." In other words, they don't have skin in the game.

It's also true that freight is a much more attractive business for a railroad, AFAICT. If you fuck up, you just compensate the owners; no one gets killed (most of the time). But the railroads don't have the option of just giving up passenger service, so they can make it so crappy that people leave

You can't outcompete the government?? Hello, Federal Express, UPS, and the Internet.

USPS First Class mail volume is down substantially, because no one writes letters anymore.

You're right that you can't drive them out of business, and they get a texpayer subsidy. I sold over 100 vinyl records via Discogs, and I used USPS because they have Media Mail, which is much cheaper than the private services.

I think both things can be true. Large entities, no matter private or public, suffer from inefficiency. But, private entities have more control over the allocation of their resources.

With a private enterprise, you have stakeholders at the top that can single-handedly reallocate resources to/away from endeavors as needed. Within government, that can be a lot more difficult as resource allocation is more complex with many more stakeholders and layers of bureaucracy. This is especially true in democracies where changes have to be approved by either the citizens or partisan groups within congress whose goals may be in direct contradiction to the success of your organization.

Idea is that too inefficient private org will go out of business (or at least loose market share), but an inefficient government will remain a burden for much longer time.

That can be how it works, but far from always. Consider Comcast as an example of an inefficient private business that has stuck around for a long time and looks set to continue to do so for some time yet. Conversely, publicly run businesses can be setup to run like private businesses that shutdown if they're not competing. This is especially the case if the public business co-exists with private businesses rather than replacing them.

My experience lines up with this. I've worked for large and small companies, as well as the federal govt. My impression of the govt department I was in was that it was reasonably efficient for its size; which, since it was huge, meant it was not very efficient. That said, it didn't seem notably worse than the large corporation I'd worked for. In fact they seemed quite similar to each other.

Google did not really do anything THAT different than Alta Vista - they just did it better, cleaner, and with less junk on your screen. The results were better, they obviously had a better algo, and they didnt blast you with ads. They also (from what I remember) had some early and easy to use search modifiers for the general public (like quotes).

Unfortunately now this has trended toward the opposite as they destroyed all their competition. I am hoping something like Kagi Search succeeds so we at least have more options.

Being an Alta Vista user before Google, one search with Google ended my use of Alta Vista. Yes, it was THAT different.

Th difference was that Google didn't optimize search to sell ads, which is the opposite of your original comment.

Is there a search engine today that, in your view, provides better search results than Google? It sounds like you’re implying that.

Google's results (in 1999) were relevant, the best results were on the first page (and very often in the first position), simple search terms worked (I'd long since mastered complex AltaVista search queries, and still had to dig through pages of results for relevant hits).

Hell, I remember when AltaVista first appeared --- I was at a conference where DEC were demoing their 64-bit Ultrix at the time, which was what permitted the search index to be held in memory in the first place (AV's secret to speed).

Google simply blew it out of the water.

That said, that was a very long time ago, and was a very different Google.

No, the differences were fundamental. AltaVista was architecturally very different, independent of ads.

It's a long time ago but if I recall correctly, AltaVista:

- Ran on a tiny handful of big iron machines. It was actually built as an advert for ultra-high end server hardware (from DEC I think?). It couldn't scale up index size as a result. Google used lots of cheap Linux machines with distributed algorithms, which could fit the entire index in RAM.

- Partly as a consequence, relied mostly on meta keywords rather than what was actually in the page. But site authors were SEO-ing the meta keywords like crazy and they were often irrelevant.

- Didn't have any global ranking function like PageRank.

Result was Google's ranking was way better, and they could serve way faster, and they could scale up much more easily.

> Google did not really do anything THAT different than Alta Vista

> The results were better, they obviously had a better algo

i'm with walter, i remember using altavista. google was _shockingly_ better. nobody tried google and said "bah i'm sticking with altavista".

> Unfortunately now this has trended toward the opposite as they destroyed all their competition

i certainly agree that they've deteriorated.

The only thing I missed about Altavista was their desktop search app for Windows. This was long before search got built into the OS.

If anything, AltaVista's query modifiers and logic were at least as sophisticated if not more so than Google's.

Google's ranking was so much better that you simply didn't need the complex queries to get vastly superior results.

Over the intervening 23 years or so, Google's lost significant ground on both factors --- its ranking no longer produces high-relevance results, and it's self-sabotaged its own syntax weaking that angle as well.

On top of my head, a couple different ways of doing search I've been using lately:

* Business search via maps service.

* "How To" search via video service.

Funny enough, best services to do so are Google Maps and YouTube.

I think TikTok is up there and possibly surpassed YouTube for the younger generation.

TikTok's user base is huge and their app makes it incredibly easy for anyone to upload and edit videos, so naturally you get lot of how to do X.

You are probably very correct! I don't use TikTok, but I noticed TikTok second-hand, via YouTube's heavily promoted Shorts.

depends on the how-to, if it's how to not ruin a dress shirt in the washing machine sure. If it's a technical in depth subject Youtube is the place to be, you can't learn systems programming on your smart phone in a minute.

I read article about how some young people are using tiktok to search for stuff instead of google. If they want to check out a restaurant or an event they would see if someone made a tiktok of it.

Google already lost the "how-to" competition.

* If you "how-to" a recipe here's a 3000 word blog post with a dozen interweaved videos and a life story at the beginning.

* If you want a "how-to" Youtube video here's a 10:01 video because gosh darn we need to hit that time count to maximize ad revenue, don't forget to smash the like button it really helps my channel

A 12 second TikTok short or searching reddit via a third party app are both terrible alternatives that are nonetheless vastly better then Google's current offering for some meaningful searches.

"Free market" does not mean a market without any regulatory structure. A market in which monopolies have formed is not free, as monopolies can force conditions on transactions that would not exist if multiple independent and competing parties were able to participate on the supply side.

Monopolies only form in a free market when the government enforces them. Like the trash collection monopoly in your city.

> Monopolies only form in a free market when the government enforces them.

This is incorrect. There are many industries that lead to natural monopolies in the absence of government breaking them up. Anything with incredibly high fixed entry costs, few substitute opportunities and low variable costs tend toward monopolies (at least regionally).

Name one.

The quintessential example is railroads, but also most utility providers (electricity, water, natural gas). Toll Roads / Bridges (often the substitutes are not good alternatives). Telecoms (to a lesser degree and declining thanks to the entry of new substitutes).

Except that railroads compete with each other, and with all the other forms of transportation.

Telecoms only had a monopoly when the government enforced it (long distance AT&T). Phone companies often had local monopolies because cities would enforce them.

People can (and did) build more roads and bridges if someone was charging monopoly rents.

Natural monopoly != zero competition. I’d recommend you read this for more info: https://en.wikipedia.org/wiki/Natural_monopoly

Yeah, I've read it. Didn't find it compelling.

https://news.ycombinator.com/item?id=32762714

What I've come to find an interesting element of monopolies is that virtually all that I can think of seem to be fundamentally networked entities. That might be physical (railroads, utilities, communications, broadcast) or logical (marketing, wholesale, retail), and might involve edges/links (rail, comms), vertices/nodes (ports, datacentres), or quite often a mix of both.

All tend to see some large realm over which there is an increasing return to scale (if there weren't, monopolies wouldn't form), and quite often a single entity may be established over multiple networks or modalities (e.g, Google controlling search, online advertising, and Web browser development).

I've been hard pressed to come up with examples of monopolies which aren't describable as networks, though that description may not at first be obvious.

Non-government-sanctioned monopolies exist as (illegal) drugs cartels, criminal syndicates, and warlord hierarchies, in which growth and maintenance of the network is typically obtained through direct (and non-legally sanctioned) force of arms, coercion, or intimidation.

> There are many industries that lead to natural monopolies

Care to provide a list of these industries conducive to natural monopolies that have monopolies broken up by government?

See my response in a sibling thread for a partial list of industries conducive to natural monopolies. Governments have had to step in to regulate each of these industries to varying degrees to rein in their power because of the fact that the industries tended toward monopoly.

Please name a company, not an industry.

That's not the only way monopolies can form, but it's the easiest one for sure.

My counterpoint is that without strong independent institutions preventing them, the companies themselves will enforce the monopoly using private security or police forces. If the government is too weak, the company will buy it. If there is no government, the company will form it's own government.

Like company towns- that's the natural end state of an unregulated market. The biggest company in the area can just buy everything and make their own laws.

Any modern examples? It seems cheaper and more effective to capture regulators—that’s what’s been happening in practice.

Free markets require a government - to prevent use of private police forces to enforce their business by violating the rights of others.

I'd be interested in knowing what definition of "monopoly" you're proposing.

Under US regulation:

"Monopolization Defined"

The antitrust laws prohibit conduct by a single firm that unreasonably restrains competition by creating or maintaining monopoly power. Most Section 2 claims involve the conduct of a firm with a leading market position, although Section 2 of the Sherman Act also bans attempts to monopolize and conspiracies to monopolize. As a first step, courts ask if the firm has "monopoly power" in any market. This requires in-depth study of the products sold by the leading firm, and any alternative products consumers may turn to if the firm attempted to raise prices. Then courts ask if that leading position was gained or maintained through improper conduct—that is, something other than merely having a better product, superior management or historic accident. Here courts evaluate the anticompetitive effects of the conduct and its procompetitive justifications.

"Market Power"

Courts do not require a literal monopoly before applying rules for single firm conduct; that term is used as shorthand for a firm with significant and durable market power — that is, the long term ability to raise price or exclude competitors. That is how that term is used here: a "monopolist" is a firm with significant and durable market power.

<https://www.ftc.gov/tips-advice/competition-guidance/guide-a...>

This article would have nothing to do with that thesis.

We're supposedly heading into a downturn, investors want to hear about 'efficiency'.

Sundar is playing the game.

And possibly the occasion to make a few cuts maybe he's wanted to make for a while.

competition (winner takes all) naturally results in oligopoly or monopoly in any market. The market is unstable.

You can fight this by unending regulation/antitrust. Or by finding better ways to distribute scarce resources.

> competition (winner takes all) naturally results in oligopoly or monopoly in any market.

Naturally? Name one. Google/Apple/Amazon/Netflix/Tesla don't have monopolies.

To name two which have formed oligopolies:

- cellphone carriers in the US

- airlines in the US

Not monopolies.

So? The post mentioned monopolies _AND_ oligopolies. Whether they're monopolies is thus irrelevant.

look at market shares over time. initially you have innovation. then many entrants fight for it. then one wins.

google is search monopoly

amazon has ecommerce and AWS.

netflix was innovator, middle period.

apple is taking 80% of profits in mobile. middle period.

tesla/electric cars is in beginning.

None of those are monopolies.

How is Google not a monopoly? Their nearest competitor is probably 10 times smaller.

bing.com

2.5% market share. Not a real competitor.

I use it. Maybe you can try it for a while and see if you find its results competitive.

Even if the search results are good, Google is still a monopoly.

You need to get the idea out of your head that "monopoly" means literally has no competitor.

Bingo.

<https://news.ycombinator.com/item?id=32762714>

Does Google not have a monopoly on search?

bing.com

"In any market" is too strong a statement. In markets with high barriers to entry or other factors that make it less competitive, sure, and we call them natural monopolies. But you're claiming that every market is a natural monopoly.

yes some markets take longer times and sometime tech changes so we never reach the end. but the end result is always the same as long as firms are involved. It's a feature of capitalism.

It's not some immutable law of capitalism. There is such a thing as diseconomies of scale, where firms that are too large end up being less competitive and unable to innovate as quickly. Once they reach that size they might borrow the power of government to crush competition and retain their monopoly, but at that point we may as well it call a feature of socialism, not of capitalism.

I'd be surprised if in twenty years of market forces (that is, assuming neither government antitrust action nor cronyist behavior making competitors illegal) more people were using Google for search than they are now.

If monopolies are always the end result in a free market, name one.

Reply on news.ycombinator.com