>Recent technological changes have transformed an increasing number of sectors of the economy into so-called superstars sectors, in which a small number of entrepreneurs or professionals distribute their output widely to the rest of the economy. Examples include the high-tech sector, sports, the music industry, management, fnance, etc. As a result, these superstars reap enormous rewards, whereas the rest of the workforce lags behind. We describe superstars as arising from digital innovations, whicih replace a fraction of the tasks in production with
information technology that requires a fxed cost but can be reproduced at zero marginal cost. This generates a form of increasing returns to scale. To the extent that the digital innovations are excludable, it also provides the innovator with market power. Our paper studies the implications of superstar technologies for factor shares, for inequality
and for the effciency properties of the superstar economy.
Yes. At this point, recommending that a 9-5 programmer quit their job to work on an indie game is equivalent to suggesting a PR writer quit their job to write the next Great American Novel.
It was always like that though. Game programming had always been the worst field for programmers, and the industry had and is exploiting the fact that some people like games enough to work on them under horrible conditions.
I don’t think it’s the equivalent to writing the great American novel though, because programming is just a minor part of game creating and arguably one of the least important. Design I’d the most important.
The author of this article rants about the swarm of unity games, but some of those unity games are better than what the author made exactly because they took design seriously.
> The author of this article rants about the swarm of unity games, but some of those unity games are better than what the author made exactly because they took design seriously.
Did we read the same article?
Not only the total number of games, but the rate of their release seems to be geometrically increasing! Holy crap. And while many of them are Unity shovelware, etc., many are polished games that a lot of effort went into. A tiny percentage are hits, but most are forgotten in the deluge.
To me that reads as someone that's fully aware that some people are putting a lot of effort in and making good games (Unity or not), but is upset by the realization that quality doesn't seem to be nearly enough. The problem is not (just) that a lot of crap is being released, but that a lot of everything is being released, good and bad. So much so that even the good things can't make good money because supply has so far outstripped demand.
It's sort of like the Netflix queue problem. I'm continuously adding things to my Netflix queue that look interesting, but my time to actually watch them is such that my chance of getting through even a majority of the queue is almost nil.
To me it reads as someone who disregards unity games, while praising well made indie games in general. And in the light of the rest of the article, the author seem to favor people who did good programming, but the truth is that you don’t need to be a good programmer to make good games.
I think supply is an issue, but I also think the author added to the problem by releasing a game that doesn’t have appealing graphics, gameplay or sound. Where as many much of the unity “shovelware” is exactly the opposite.
> Where as many much of the unity “shovelware” is exactly the opposite.
I think the difference of interpretation we're having is that I think you are interpreting "Unity shovelware" to mean "If it's Unity it's crap", which nobody who follows that space could easily defend (a lot of high quality well known games and publishers use Unity). But a lot of shovelware uses Unity, because it's easy to get assets for and publish with.
It's sort of like saying "Java enterprise crap-ware". I wouldn't assume that means all Java programs are crap, or that all enterprise software is crap, but that of the crappy software targeted towards the enterprise, a lot uses Java. That's not an indictment of Java, and might actually be the opposite, given that it has qualities that cover up other poor choices.
Shovelware typically does not have "appealing graphics, gameplay or sound". It is term used for low effort games and imply nothing special in all aspects.
Shovelware can have appealing graphics and sound by way of licensing premade assets, but you can't really buy a game design and paste that into your own title to get appealing gameplay.
I get the shovelware feeling from Netflix too tbh; as with the Unity games mentioned, there's a lot of effort and they're high quality, great writing, visuals, direction, they tick all the boxes and we live in a golden age of TV series and films - but there's just so much of it. Not hearing about any of the shows or movies beforehand from my social circles doesn't help either - the network effect is super important. And that's from a service with a fixed fee per month, so the bar for watching something new is super low.
Re. The shovelware bit: Once upon a time, when men were real men, etc, making a game meant writing it from the ground up, writing your own engine and then writing the game logic on top of that engine to make something unique.
Now that game engines are commonplace, free and work better than what any single developer would be able to come up with after a lifetime of hard work the end result is that games have a very hard time to differentiate themselves from each other. There isn't really an unlimited space of game scenarios out there, even in the days of the 2D arcade games after a couple of years it became much harder to come up with something truly unique.
It was never difficult to "come up with something truly unique".
Your unique thing might be rubbish, but that's not the problem for the millions of shovelware Unity asset flips.
Jim Sterling has covered this at length, even running a competition to show that people can take a horribly over-used cheap asset and do great original stuff - if they try. Asset flips don't even try.
The are tens of thousands of games in which you are an elite soldier running around shooting zombies. There are zero games where you're a pot plant using psychic powers to create sculptures.
Like Hollywood the video game industry chases trends until they're beaten completely to death, and then goes one more round just to make really sure.
I think there has been some windows of opportunity for single-developers teams.
- In the 80's and 90's when PC games where simple enough.
- At the beginning of Steam
- At the beginning of the App Store and the Play Store
At those time, there wasn't that much competition so while it was still hit or miss, you still had a decent chance to be successful as a single-person indie developer.
I remember buying games in the 80ies and 90ies, I’d go to a supermarket and rummage through a mixed box with literal hundreds of different $5 games and pick the one that looked the most exciting.
The biggest treasure I found that way was the original x-com.
I don’t think times have changed that much. The upcoming game I’m most excited by is it lurks below, and that’s being developed by a single developer. Stardew Valley released in 2016, after the explosion. Into the breach, though not single developer did well in 2018.
I mean, it’s not uncommon for solo-small teams to top the steam charts, but you really do need both quality and a little luck. I just don’t think that’s different from how it’s always been, and I think the author of this article in particular lacks quality.
What has changed though is the amount of games people own. In the 90ies you didn’t have a backlog of thousands of games that you picked up from a humble bundle where you only really wanted one game. You also didn’t have MMOs or forthnite competing for your attention.
So I agree with you somewhat, it’s become much harder to sell bad-mediocre games in the past few years, and that’s 99% of indie games. And there is obviously always a market for remakes when a new platform/generation arrives.
Novels are their own problem, but it's a similar one; there's hundreds of thousands of great writers and publishing is trivial nowadays (iirc the longest work of fiction ever made is a Super Smash Bros fanfiction published for free), but getting 'mindshare' (or, just some attention - and that's before making a sale still) is hugely difficult.
Is there really a difference? I know an acquaintance that published a novel they worked on for a few years. They didn't share any numbers with me but they were clear, they made no money. I assume there are thousands of people trying to sell their novels, and a tiny minority succeed in making (big) money.
> and a tiny minority succeed in making (big) money.
Who is making big money JUST off novels? Not speaking, not tv deals, writing.
Meanwhile I know several game developers making a living. Setting out to do it still follows a success distribution, but it’s fundementally achievable in a way unavailable to writers entirely.
They exist. But those writers are making money are doing quantity over quality. I.e. they're releasing like 12 pulp novels a year that they churn out once per month, and build up a fanbase over time with every new release, who then buy up their quickly growing backlog of novels.
Seems like urban supernatural or romance (or urban supernatural romance) seems to be the main genres where this is working. My girlfriend follows several writers who make a living this way, and is wanting to give it a try herself, that's how I'm aware of it.
You can't write The Great American Novel or spend years writing your novel and make money this way though. It pretty much requires the momentum and consistent quick releases for it to work. You're making sugary garbage that's consumed quickly and forgotten just as quickly.
I'd think this is more true for a writer than a programmer. You could probably attempt to program a game by spaghetti coding, copy pasting from stack overflow and tutorials, not using version control or tests, and working solo - at the end you'd have a lot of bad habits and not so much that would be useful at a professional studio.
Some people use mistakes as a launch pad to learn a better way. Not always, and not always immediately, but sometimes it takes dealing with the negative consequences to make the better solution obvious when finally encountered.
Sometimes convincing someone to spend the extra 10-20% of time or effort to do something "the hard way" is nigh impossible until they've spend time doing the alternative and know the extended pain that sometimes results.
It's sort of like seat belts. When they were first required in all vehicles, many people still didn't bother with them. Even before cultural indoctrination took over, a lot of people eventually started too. It took some close calls for people or their friends and family for them to finally make the effort, and such a small effort it was. I saw exactly this play out with my own parents and their siblings.
Game development is hard, it seems like it's completely hit or miss, there is no middle ground. I've released 5 games, but I have to say, spending 2,600+ hours on something without doing market research or at least starting small seems crazy.
My most successful game was an app for iOS which made me a grand total of around $30. The difference is I only spent around 80 hours working on it (40 or so for the game, 20 or so on the level editor, and 20 or so making levels). I had no money for marketing, had done no research beforehand, so I pretty much knew it was going to fail but did it as a fun side project to learn some new stuff and for my CV.
Compare that to the various website side projects I've worked on over the years which have made me hundreds of thousands of dollars (over 10 years). Game dev sounds fun, but if you're looking to make money, I would definitely stay away from it.
Instead of making a game for some category on Steam and hoping it sells enough in the first month before sales drop off, I wonder if it could work to make a super-super niche game where the goal would be to rank for a Google search term instead.
You'd try to be very long-lived, steadily selling for years, say just a few copies a week. All traffic would come from people who have some incredibly niche interest they are searching for, say baptism planners discovering your "baptism planner simulator 2000".
If you continued to make games like these say one per month or two, it might be possible to gradually build up a liveable income from them.
I guarantee that if you drop the "planner" that game would sell like hotcakes as soon as a medium sized YouTuber/Twitch streamer notices it.
There is no way a game about dunking babies doesn't get enough shock value to get a decent return on investment, especially if it has wonky physics, "bad" controls or other emergent gameplay.
For a deeper game, maybe bring back the "planner" aspect to add a simulation layer on top. Make a tycoon game where you manage a church and the baptisms are just one of a few minigames.
Would you say that the difference between the games side projects vs the website side projects are the mindset in which you have executed those projects?
Not really. The market is large and fairly mature, it's doing fine and continuing to grow. The video game crash is titled that for a reason, the market shrunk by 97% in a short period.
There are a large number of amateur creators who are trying to win the lottery (either from a fame or monetary perspective), and most won't and will see little recognition and remain obscure or very niche.
This is seen in basically every area of media. Music, books, movies, etc and is perfectly normal.
>In 1986, Nintendo president Hiroshi Yamauchi noted that "Atari collapsed because they gave too much freedom to third-party developers and the market was swamped with rubbish games". In response, Nintendo limited the number of titles that third-party developers could release for their system each year, and promoted its "Seal of Quality", which it allowed to be used on games and peripherals by publishers that met Nintendo's quality standards.
In 2017, 21 games per day were released on Steam and most of them were trash.
The Atari crash was more akin to the .com bubble. Vaporware and no way to tell what was in the bag until you bought it. Video games are a giant mature market dominated by big companies with sophisticated business models. Just like web businesses these days.
I’d say sports, too. TV made it so some players were playing for hundreds of thousands of people nationally while others still entertained only locally.
It’s also in the process of starting to hit indie development in general and startups. It’s an interesting process but life for your typical indie or startup SaaS is going to get as tough as life for the typical “App Developer” got a few years ago.
The problem with games is the amount of effort involved just to determine if it is “fun”. When I was in it cancelling a game 1 year deep with 10 people was normal.
I think this is why minigame-style games (battle royale, fighting games, first person shooter competitive, etc) are so common. You can implement a simple version of your game idea in a weekend and, if your idea was good, people would likely already have fun playing it.
Never really did modern game development, but can you really implement a first person shooter with a twist over a weekend? How would you go about it, create a mod for an existing game?
Almost no one writes their own FPS entirely from scratch in the past decade outside of a handful truly enormous industry players, much as no one really writes the entirety of a web app themselves.
You choose one of the many existing game development frameworks and rapidly prototype a concept, if it works you keep iterating on it etc etc. This works in games as well as any other major software endeavour.
The matter of having the talent to produce a new idea that’s actually fun to play and marketing it successfully is arguably the hard bit here, rather than the technical implementation details. I’d make the argument today that the technical barriers to entry for game development are almost _too_ low now, which is why partly why Steam is filled with so much “indie” junk.
Asset stores allow you to have placeholder mechanics like you used to have placeholder art. So placeholder camera, controller, collision detection, and a placeholder level. Then you get to rig what you want to test on top of that, or just sell it as is which is called an "asset flip".
You could do that or a frramework like Unity takes care of lot of stuff for you.. weekend maybe a bit extreme, but you can get something running and functional in very short time
You can download Unity, purchase and install for example UFPS that will handle most FPS functionality for you (includes basic gun etc model) and you can use Unity networking for LAN connectivity.
Let's say the twist is a Battle Royale game, you just make a Unity terrain and a basic sphere that becomes smaller.
After that coding the extra bits such as HP, HP loss due to sphere etc isn't a big deal.
That's one way, doing it via a mod also works, but you really need knowledge of the product you're modding (although the same is true of e.g. Unity, if its your first week with it it won't be that easy).
What you are saying is true. However, throwing your hands up in the air and saying “I guess I didn’t get lucky” isn’t quite the correct response either. Success is a matter of exposure and conversion rate.
1,016 visits, many of them random and untargeted, is simply not enough to make a statistically significant decision about the viability of this game. Conversion rates for any product tend to vary wildly within specific demographics. Some games perform abysmally with wide audiences, but may have very high conversion rates with specific, well defined groups.
To be sure, there are games that don’t need to look for their niche audience. They become viral sensations because they appeal to mass audiences. They get enormous amounts of “earned media” - viral clicks from people talking about and uploading video of themselves playing - and those are the games we hear about and consider to be “Superstars”. But that in no way means that games that aggressively target some relatively small group of people that the game actually appeals to with paid advertising cannot be very financially successful. Maybe not billion dollar blockbusters, but I’d imagine this author would be happy with a six figure income from his work, which is entirely possible if he finds the right audience and applies the same work ethic he did to developing the game to marketing it.
He needs to figure out who likes the game, what makes them like it, and then use the plethora of online ad platforms and targeting options to find more people like them. His game is not a steaming pile of crap, so he will find a paying audience for it if he looks.
No, because a "winner take all" industry is one that trends towards monopoly, that's not the same thing. No one person will ever have the entire video game pie.
"Superstar industries" follow extreme power law distributions, where, due to the low barrier to entry, high ceilings, and limited consumer base (there are only so many hours people can spend on entertainment), the pie isn't growing and new entrants aren't going to be able to carve out much that hasn't already been claimed.
It's not a new thing, the name comes from the music industry which acts the same way. Anybody can record a song and print it on CDs, but the market is already saturated: you have to either really really really stand out like nobody has in decades (An event so rare I couldn't find any examples), or you have to get the support of one of the big players (game studios or music labels) to lend you their resources and audience.
Nearest example I can think of, mySpace was where she got noticed so that dates her breakthrough to a good decade ago.
Adele was legitimate however others have pretended to have an Adele grade story. Lily Allen also claimed to have been discovered on mySpace but her dad was rock and roll celebrity so her efforts can be dismissed as nepotism. Often these links to rock and roll celebrity are not obvious because performers use stage names that sound like real names. Their children don't use their parents fake surnames.
Adele wasn't self-published though, right? I'm not super familiar, but based on the wikipedia page she got picked up by an XL Recordings talent scout who put her on other albums associated with that label, including appearances on BBC, before even her first album, which had 32 credits.
Being really good and self publishing helps you build your resume in a superstar industry, but it's nearly impossible to actually achieve commercial success without a well-established sponsor. Adele probably wouldn't have gotten far on her own if XL or some other label hadn't picked her up. Her accomplishment before then was a really good resume.
Her mySpace tunes were self published, however, getting XL Recordings on board has parallels with startup culture. It is like the importance of having a co-founder, proof that you have convinced someone to back you, then with that, you can scale things up.
XL Recordings came from the rave scene, the Prodigy were their first big earning group, however, for every Prodigy there were scores of 'Dome Patrol' grade releases that nobody except for DJs ever heard of. XL moved on when the rave scene died to other acts in different genres. They 'pivoted' to use the parlance.
I know Adele worked with a label however it was a genuine thing her getting spotted on mySpace and not some fake back story, as per the Lily Allen example. I don't believe music executives really would listen if I posted my singing efforts to the internets, or if anyone here did likewise. Yet that did happen with Adele.
I'm not saying that Adele's story isn't remarkable or that she isn't talented, but that she didn't break the industry pattern, finding her own slice of the pie without help. XL already had audiences, resources, collaborators, and connections, and a production pipeline that she was inducted into. Her debut product took 32 people to make, on XL's dime. It was not a Stardew-Valley situation.
Justin Bieber was discovered through YouTube. And he helped contribute to Carly Ray Jepsen's career because he saw one of her songs on YouTube, covered it, and helped it go viral. The Weeknd used YouTube at the start of his career as well.
I don't think this phenomenon is limited to one form of media. "The Martian" by Andy Weir was originally published on his website one chapter at a time. The internet has, in general, lowered the bar to publishing any kind of work significantly - be it games on Steam, songs on YouTube, or ebooks on Amazon. There are now new avenues into the wider industry, but the big players are still the big players.
People break through all the time. But there's no formula for it. It's luck + talent + zeitgeist + social connection. You don't need all 3, but it helps a whole lot. And you can have a whole lot of any of them and still not have lasting success. This is how the music industry has been for a long time, although the dynamics are always, well, dynamic.
Indeed, you have to look for niches -- and find the ones that are in
1) high demand (not just high demand in general -- high demand for new/better content by a new entrant);
2) low supply;
3) well paying customers.
I'm sure despite the flood those will still exist, as long as there's still a good amount of differentiation between people's preferences. A niche is a small subset of in the space of products that a small subset of customers have strong preference to. This needs preference differentiation.
This preference difference could be toward many features: cultural setting, story style, gameplay style, aesthetic style, etc.
There's perhaps a combinatorial advantage here: if sensitivity varies significantly across many of those parameters, just choose any unexplored subset, optimizing for 1/2/3 -- there are exponentially many in the number of distinguishing features (if you browse Steam tags you see this is potentially a very large number!).
Making the games you want to play is actually not great advice. You need to find a niche and hit it hard.
Maybe that used to be metroidvania, but that's long over. Visual novels have been on the rise for a while, so are rougelikes and card games, but those are saturated now. You want to get in before the big streamers and reddit frontpage and ride the wave.
Who knew a Harvest Moon clone would be the biggest indie success of its year? Maybe we can systematically analyze old games to identify overlooked gems and valuable revivals.
>Making the games you want to play is actually not great advice.
I don't think it's bad advice, though. If you treat genre and its elements as only means to an end, you wind up with exactly the trends that oversaturate the indie market now, because everyone is trying to "find a niche and hit it hard." Games are entertainment.
I think you should at least learn to want to play the game you're making. At the very least, you should know why the customer would want to play it. Ticking off boxes hoping to find combinatorial success isn't enough... that "battle royale farming simulator" still needs to be entertaining, or at least capture interest.
> Making the games you want to play is actually not great advice. You need to find a niche and hit it hard.
I'd say you need both: you need to want to do a game you'd want to play, because you'd have an easier time recognizing if you're doing something that would appeal to at least one person. And if it's in a niche, you'll have a better chance to get coverage and buyers. But if you just try to hit a niche hard without knowing what makes the games in that niche good, you'll have a hard time making something that'll interest players.
I disagree with that, but it's just my opinion. I think artists must move beyond their own taste. A sushi chef doesn't really have to like squid.
I think using your own taste is a crutch. If you get good at it, you should be able to know it's good regardless of your own feelings.
We see this when game developers are shuffled around and work on ones that are completely different from their passion projects. The good ones still excell.
Even if one is good enough to recognise a good game regardless of one's feelings, I'd say it would be harder to work on a game type that doesn't appeal the artist. I hope that even if a sushi chef doesn't like squid, he like some of the sushi he'll be making.
But you are actually competing with the whole developed world. There may be localised tastes or patterns to follow but your competition is from the whole world.
Sure but I think this really applies to industries under a free market in general however the dynamics in the tech sector are such that a single person with little to no startup capital can create something with paranormal returns and that is really what indies are after (perhaps after having their idea come to life). I wonder if there any other industries that have this same allure
I don't think that it's just technological change which has brought us there. It's also because of the last 10 years or so of Federal Reserve Bank policies. The elites have figured out how to maximise their profits during expansionary economic times and a big part of the elite strategy is limiting the number of winners.
It's pretty clear that a lot of jobs today are useless and if you've worked for enough big companies, you might have realized that big companies are extremely inefficient.
If big companies are so inefficent, why then are they able to generate so much profit from the economy?
The reason is monetary policy. The Fed constantly 'prints' new money out of thin air and injects it into the economy via a combination of 1. loans to the government and 2. open market operations.
Guess who gets the first dips into these two huge pots of newly created money?
In the case of government money (loaned by the Fed), big corporations get a huge chunk of that money pot because they get all the lucrative government contracts.
In the case of open market operations; big finance firms like Goldman Sachs and big traders with insider information are the first to get their hands into that pot.
Basically that money which is meant to 'trickle down' isn't actually doing that; instead it appears to be trickling back up to government officials who are enacting policies that benefit corporations and increased centralization of wealth.
>> New money that is anticipated will lead to anticipated price rises. So it doesn't matter (too much) who gets first dibs.
There is a significant delay before the newly created money affects wages.
Shareholders benefit from the new money instantly (since the market reacts to it quickly and it drives up the value of their stocks) but wage earners won't benefit from the new money until they get their annual 2% salary raise at the end of the year.
That delay is significant because it allows shareholders to compound their ROIs several times before wages catch up.
> If big companies are so inefficent, why then are they able to generate so much profit from the economy?
On the one hand, no one has yet figured out how eliminate inefficiences out of large organizations. They're all like that, usually the bigger the more inefficient (the government, because of its size, being obviously the worst offender).
However, it seems that it just takes dozens (if not hundreds) of thousands of employees to run say a top tier car or microprocessor company. You just can't do it with less, so the inefficiencies are the cost of producing such advanced stuff as say BMWs or Intel processors.
The currencies of developed countries are pegged to the US dollar and so their reserve banks are just copying whatever the Fed is doing; their main goal is to make sure that the value of their currencies doesn't change too much relative to the USD. All countries reinforce the same economic distortions on a global scale.
Absolutely. All things are turning into the economics of superstars so long as they're subject to network effects and the widest, most fluid and well-informed possible market.
Uber drivers will turn into the economics of superstars. Amazon warehouse workers will turn into the economics of superstars. You'll keep on getting individual ultra-high performers who dominate whatever area it is, and more ordinary performers will just plain fail.
I'm not sure how your conclusion follows. Uber drivers and warehouse workers are unlikely to undergo this effect because no matter how good an Uber driver is, they can't be in multiple places at once. Similarly, there's a limit to how much one warehouse worker can do.
Kid of. You'll still see a small minority of Uber drivers making money and a large percentage breaking even or losing money. It's like that in the real-estate market, where a small minority of agents take in most of the profit and a large number breaking even or losing money. This is also why you see trade groups frequently supporting policies to increase barriers for market entry. In real-estate this means higher and higher licensing standards. For example in Canada it costs around $6k and 3 years to get fully licensed [1].. which is crazy. But the real-estate lobby wants it even higher.
Real estate is infrequent high value transactions with high margin/commission. Doubling your real estate sales can well make you rich; doubling your Uber driving would not.
No doubt there are differences between the markets but at a high-level similar forces act on both. That is, in both cases, the vast majority of participants either break-even or lose money. A small minority makes money (and therefore takes in most of the profit to be had). The uber market, because it centralized and has an impartial dispatcher (i.e. Uber corporate), that market will probably be more egalitarian. Overall though, the same dynamics play out.
For real-estate, some numbers that I saw had 51% of licensed real-estate agents making one or zero transaction in the past year.
Amazon warehouse workers are limited by human physiology to move boxes within 10x of the average box moving speed. That disqualifies them from the superstar economy.
The Macroeconomics of Superstars[1]
>Abstract
>Recent technological changes have transformed an increasing number of sectors of the economy into so-called superstars sectors, in which a small number of entrepreneurs or professionals distribute their output widely to the rest of the economy. Examples include the high-tech sector, sports, the music industry, management, fnance, etc. As a result, these superstars reap enormous rewards, whereas the rest of the workforce lags behind. We describe superstars as arising from digital innovations, whicih replace a fraction of the tasks in production with information technology that requires a fxed cost but can be reproduced at zero marginal cost. This generates a form of increasing returns to scale. To the extent that the digital innovations are excludable, it also provides the innovator with market power. Our paper studies the implications of superstar technologies for factor shares, for inequality and for the effciency properties of the superstar economy.
[1] The Macroeconomics of Superstars, Anton Korinek Johns Hopkins and NBER, Ding Xuan Ng Johns Hopkins, November 2017 https://www.imf.org/~/media/Files/Conferences/2017-stats-for...
[2] The Economics of Superstars The American Economic Review , Vol. 71, No. 5. (Dec., 1981), pp. 845-858. http://www.uvm.edu/pdodds/files/papers/others/1981/rosen1981...