Not sure that there's any paradox here. The rise of a freely-accessible network connecting all humans (i.e. the Internet) has just made it such that hundreds of years of cleverly and carefully assembled middle-middlemen that were required to keep the economy going are no longer required. This affects the software ecosystem that existed right before the Internet but impacts everything else as well: taxis, hotels, etc.
Moving forward, assume that middle-men are gone, unless required by legislation or differences in sovereignty (i.e. still effectively legislation). The net result is that it's cheap to build and procure yourself practically almost anything, with the surface of things you can't build/procure yourself increasingly diminishing given the network effect of large swaths of the human race rushing to the Internet to build ever increasingly sophisticated modular components with each trying to outflank each other with a lower price/barrier-to-entry.
If we can 3D print/carve materials we can buy in bulk in a cheap fashion from freely-available software/data, we are likely looking at a fundamentally different economy than the one we are in today.
Wishful thinking. In the real world, middle man competition is being replaced with middle man monopolies, e.g. Amazon between manufacturers and consumers, Apple/Google between app creators and users, Google between websites and readers, Uber between cabbies and travelers.
I believe it then becomes a matter of perspective. A single middle man will be a monopoly to one party and a monopsony to the other. So for consumers (and regulators), this state of affairs would be indistinguishable from a traditional monopoly.
I'm not familiar with the associated ancient Greek word for middleman (-poly and -psony come from ancient Greek), but in modern Greek "μεσαζοντας" (pronounced meh-sa-zon-das) is "middleman". So that would suggest something like "monomesazontia".
There is also "μεσίτης" (meh-see-tees) which means broker / facilitator. That would suggest "monomesitia".
But if you're willing to mix Greek prefixes with English/Germanic roots for "dealer", then I like the sound of "monomongery" :)
The middlemen won't be done because they offer advantages many companies want. They handle so much of the sales, distribution, international issues, and so on that it grows the company. Not as profitable as direct sales but it's hard to get them.
Microsoft's OEM deals and Oracle's sales team drove quite a bit of their revenue. Never underestimate power of good middle men. Just make sure you get a good deal and preferrably have a transition to direct in the future.
The title is a half-truth: commercial software is certainly in the decline while also being a huge, profitable business. The SaaS and ad-driven vendors get most of the attention. They might even be most of the startups. Yet, there's a steady stream of companies building software that's worth paying for and making money on it. On top of that, there's always the "improve legacy systems" market where you makes something they can't get away from even better. Solutions building on Microsoft, Oracle, mainframe, AS/400, VMS, and so on all come to mind. They stay around, so your enhancement might stay around too.
The two ensure commercial software won't go away or even be marginal in terms of where the profit is. SaaS is a race-to-the-bottom due to intense competitiveness. Ad-driven model can be lucrative but is also high risk. Building a product with strong lock-in effects is the only method proven to last decades. From there, you can decide whether you want the risk of the ad model (Facebook) or the profit of commercial model (Microsoft, Oracle). One looks more appealing. ;)
[Disclosure: I'm the author] No argument that up-front, perpetually licensed software is currently a huge, profitable business. Windows and Office alone, as acknowledged, generate essentially $44B by themselves. There are also profitable businesses that pursue the traditional perpetual license model - Palantir being perhaps the most notable example (although from conversations with employees there, they are increasingly being pushed towards alternative models).
But the point of the Software Paradox is not that companies are not currently and cannot in future make money from software, and that there are no exceptions to the rule, it's that the trajectory broadly is not promising. Looking across a variety of software categories, from mobile to operating systems to infrastructure to tooling to consumer, the trendline is downward in terms of their up-front realizable revenue potential.
None of which should be taken to mean that software is going away. If anything it's becoming more important, hence the use of paradox. But developers and companies seeking to monetize software should at least be aware of the market context which is that it's becoming more difficult to make money from it the way that it was possible to even a few years ago. Even the lock-in mechanism, which is correctly cited as a method proven to monetize customers efficiently, is more easily replicated in services businesses (e.g. cloud) than it is in traditional on premise software, where open source has become both an expectation and a means of combatting lock-in.
If one looks at all of the available evidence and decides to proceed with a traditional software licensing approach in spite of the observable challenges, it is certainly not impossible for them to a) generate revenue and b) be profitable. But the degree of difficulty attached to this model has gone up considerably, and looks to only be getting worse. And this isn't just an academic theory, it's something I hear almost daily in conversations with vendors.
Appreciate your clarification. I agree that the overall trajectory is not promising in the mass market. What are your thoughts on the pay more for quality or security market for software/systems? Have those analyzing the data made this distinction to see how much the trend affects that market segment? Many that I know of still use the old model's of licensing or expensive appliances although the startups are doing things with new models. I think, if that niche had same effect, it would reinforce your claim even more. If it wasn't as impacted, it would give the best spot to invest in and market apps with profitable, licensing models. What I often called the Toyota Strategy.
"Even the lock-in mechanism, which is correctly cited as a method proven to monetize customers efficiently, is more easily replicated in services businesses (e.g. cloud) than it is in traditional on premise software, where open source has become both an expectation and a means of combatting lock-in."
I'm actually starting to agree with this more and more as I look into how they design it. Facebook is the perfect example: containing people's whole life with no easy way to move it. The cloud vendors usually allow unlimited incoming data and charge for outgoing. Tricks like that combined with low, entry barrier might make for nice lock-in over long term.
"But the degree of difficulty attached to this model has gone up considerably, and looks to only be getting worse. And this isn't just an academic theory, it's something I hear almost daily in conversations with vendors."
Oh, like I said, I buy that. Title just read like another piece on how that type of software was done altogether. The claims about the trend in the book and your comment are right on. I guess you could say that's what I really was griping about. ;)
For existing software providers, the book recommends a transition to an recurring revenue model ala SaaS. Short term business incentives and technical limitations in an existing code base, may make this a tuff sell to existing software providers.
But the long term advantages are well argued in the book (less support costs, higher long term revenue per customer) and perhaps worth the costs of transition.
The recurring revenue model is particularly advantages for products with a strong lock-in.
You have a good point and it complements my comment rather than contradicting it. I should've been clear that, by SaaS, I meant the cloud-hosted-monthly type we see the most. Traditional software that's paid for monthly is still licensed, commercial software. You are totally right that this can be financially advantageous to the company: it's the reason many of us avoided certain software in the past. ;)
Now, with the entrenched doing it, it's looking to be difficult to avoid for buyers. One negative side-effect is that it might be more difficult for newcomers to start without a lot of capital. Snagging a few customers for $1,000-3,000 each can cover costs more quickly that slowly building up subscriptions of $10-50 a month. On the other hand, it might make it easier to sell.
Still not sure on that part. Wonder if anyone has published a detailed analysis of costs and benefits of that for a startup doing a native application on subscription. Probably just missed it.
The recurring model is proven, though, for established software companies to iterate more on and squeeze more profit from their offering. The shift of the profitable firms toward that market argues heavily for it.
I'm not sure those big vendors will stay around, to me Oracle looks particularly vulnerable. They may have a lot of "locked in" customers now but all they can hope for from that strategy is a long and eventually terminal decline.
The problems facing the biggest vendors have to do with lack of innovation and adaption. For example, Oracle could've been investing in and acquiring all kinds of clever DB technology to integrate within their own stacks. They did little in that space. Now those companies are eating their potential market-share.
First Oracle acquired Peoplesoft. They could've changed their name or Orasoft. Then they acquired Siebel, and should've changed their name or Orabel.
Jokes aside, Oracle has been on a huge acquisition spree for many years. Fusion (based on the Weblogic acqusition) was supposed to integrate all the other acquisitions, as well as partner products. It was over-hyped, and took years to fully materialize. And then Oracle missed the cloud boat, and has been playing catch-up ever since.
Probably most importantly is their sales strategy. As a potential Oracle customer all you had to do was mention Microsoft, and Oracle products would get heavily discounted. And then the customer would get stung for extortionate consulting fees and license renewal costs.
They used to be a quality name in the industry, but have slowly and very steadily eroded that reputation.
I was talking about the companies marketing document, key-value, cloud, etc. databases. New & improved, at least different, technologies that were coming from many different areas and getting lots of users. They might have acquired many top-name companies in those bunch and Oracle's NoSQL products have millions of paying users. I haven't followed Oracle acquisitions closely so I had to take a guess: Oracle's culture would make them primarily acquire things similar to them: RDBMS's, ERP, etc. Then, companies doing things in different ways would keep scooping up customers that might have bought Oracle.
Was this the case or was I way off? You seem to know more about their acquisitions.
Hmm - so what would be a title that would not be a 'half-truth'?
There is only so much space in the title - you cannot load it with all available information. The article itself goes in much detail about what you describe shortly here - and what is the most surprising in it is the decline - it seems to me thus as a very good choice for the title of that article?
The title reads like it's gone or going away. It's not. The book itself supports the opposite: a much smaller marketshare with revenue coming from prior and new models with a concentration toward subscription.
I'm not the kind of persion with artistic or brevity skills you want making a title. An honest example anyway: "The Rise and Decline of Traditional, Software Companies." Or something similar that indicates the game is changing, up-front proprietary is on decline, and importantly commercial software is still a multibillion dollar business with successful startups appearing each year.
Steam engines for commercial travel had a rise and fall. Software market is just adjusting with cetain models taking a financial hit.
It's really twisting things in order to fit a wierd corporate narrative on FOSS.
"Even companies like GE are helping to fund noncommercial software, having con-
tributed $105 million to Pivotal, the home of projects like Cloud Foundry." Uh... they invested for a 10% stake in a for profit company which creates commercially focused free software, among other things. The corporate doublespeak runs deep.
Yeah, that is deceptive. Facebook is a great example of a company funding FOSS. Redhat is the canonical example. EnterpriseDB is another example. The defining characteristic is the difference between what they're taking from a project and what they're contributing. Most companies prefer a 95-100% one-way relationship in their favor.
The tl;dr: 'The economic value of software is declining even as the strategic value of software is rising.'
When I applied this framework to Docker, the magic sauce of Docker pops out in an obvious way. I had asserted in other threads that Docker's thing is packaging ... but reading this paper, I can see why Docker is moving so fast.
And go figure, it's written by Stephen O'Grady of RedMonk.
I think Docker is selling (or riding) an ecosystem, not "just software". If they don't claim major mindshare, their could be replaced by a competitor. For example, see Bocker, "Docker implemented in 100 lines of bash":
> 'The economic value of software is declining even as the strategic value of software is rising.'
Sometimes as a geek I feel a certain sympathy for 3rd world nations who happen to sit on a scare natural resource and are forced to give it away for a song when asked by merchants in military gear..
Have you ever been to India? Do you know what they would be doing to survive if they couldn't program for U.S. businesses from across the globe?
I saw people in India eating trash. I've been there. I have seen the other jobs in the area. Not everyone was dirt poor, but there were many. Meanwhile, there were also affluent people, with drivers and servants.
Trust me, the people who work to become programmers in India are among the middle to upper class.
This is not what was meant by "low economic value". Selling software itself will become increasingly untenable. What software enables will become increasingly more valuable.
Your metaphor that "you happen to sit on a scarce natural resource" is not apropos. That metaphor still treats software like an economic, rather than a strategic advantage.
If you feel like you're being exploited like someone in a sweat labor shop, then you're obviously not leveraging this changing reality to your advantage. You're probably still working at a place that calls whatever you do "IT". It's time to leave that burning ship. You're better off looking for a startup that is writing a SAAS that competes with whatever your current employer is doing. Your work (assuming it is good) will at least be respected by your peers and the leadership.
If software has not eaten the space that your employer occupies, it will soon.
People who can create enablers are only increasing in value.
The market for computer code is still growing exponentially. Selling software directly is seeing a decline, but incorporating software into other things, or delivering it as a service is on the rise.
Anybody adding value by some other means, hardware, professional services, etc... is going to be looking for and willing to pay people who can apply software to add value.
I've sat in meetings where decision to dismiss an entire floor of programmers in one of the nyc building of one bank included open and frank assessment of outsourced work ("4th iteration" "crap" ...) by one of the executives.
("Mr. MBA" is just another worker, like you and me.)
I'm confused: is the executive dismissing outsourced work as crap, or is he advocating outsourcing because the entire floor of NYC programmers' work is crap? Did they make the decision to outsource the work?
He noted that it was crap, but then again the meeting was about bringing in /more/ so the floor could be cleaned up. I'm sure this equation resolves at a higher interest level, but it did not make much sense at that table.
Looks like the linked PDF is with the "Compliments of PayPal", Google Cache[1] still has that version of the O'Reilly store page while now actually cost money again. So I guess someone was clever to save the download link while the promotion lasted? :)
From the description on the O'Reilly page I get the impression that this book is perhaps 90% hindsight and 10% helpful advice for entrepreneurs in software. Is this a correct observation?
There are lots of good lessons in history as well. Remember that trends in computing tend to be cyclic (for instance: centralization/decentralization).
Cycles are an attractive framework for thinking about technology progress, but is it true? For example, the app->Web->app apparent "cycle" happened exactly once. And there is a government jackboot on the neck of decentralization. I don't see repetition or commonality. All such supposed cycles happen on the backdrop of huge growth rates.
The 'What to do?' chapter is 12 pages out of 63 so I'd estimate nearer 20%. Suggestions include doing other stuff related to the software such as hiring taxis (Uber) or selling shiny hardware (Apple, Nest).
We launched in December 2012 our offline desktop software product for business and freelancers (niche market with a TAM of about 100,000 users) only as a subscription, no perpetual licenses available. Previously this project was freeware (closed source) so our 2.x version remained freeware and our 3.0 was only available as a subscription sold in chunks of 1 year-subscriptions as opposed to monthly).
In retrospect, we are very happy to have chosen the subscription model because it resembles a lot the freeware model: for maintenance you just focus on the latest version/build issues because 100% of your customer base is entitled to have it, you don't compete with yourself across paid versions and don't need marketing effort to explain customers why they should upgrade every year. Your sales effort is also lower because your price point is lower.
It certainly looks like there is no money to be made anymore in compilers, or software development tools in general. That market seems to have been eaten by free open source software.
I wonder if there is any advice for software tool builders? Has the money really dried up?
There's several private companies that make money selling compilers: Green Hills and PGI comes to mind immediately. Open compilers are good enough that commercial ones have to be better in general or on specific platforms to justify price. Plus, they usually integrate into other tools that do things FOSS cant or doesnt do well.
One area that's an untapped market is compilers that automatically and efficiently protect legacy code. I've seen academic and FOSS methods that do well. A thoroughly tested, supported, integrated with IDE option would probably sell well. A cross-platform even better.
But Jetbrains is making money. There's also a bunch of static analysis tool vendors.
We can model open source as cheap copycat inports: you can't make money at the low end, but must go up market. You have to keep improving to stay ahead of the tide, but eventually there's no up market left and you are sunk. So you have to jump across to the next island.
Commercial software may be getting a smaller piece of the pie, the pie is getting larger quickly.
I basically agree with Donald Knuth: open source (I would prefer it be libre) software will continue to be more important with passing decades. I expect that if the human species survives long term that in a few hundred years the world(s) will be run at partially by very old and very stable open source software.
I have had access to computers since about 1960 (thanks Dad!) and the improvements have been exponential. That said, in the distant future when computer science is a mature science, for many tasks rock solid software will likely be more important than the decade's latest bells and whistles.
It's true that free software is eating into everyone's margins, especially unsustainable business models that run entirely by burning cash and continually raising more and more money in hopes a giant will come and swoop it off it's feet.
However, I think there's a problem with attempting to take the assumptions from macro view and drilling it to individual companies and startups. Plenty of awful software with awful websites sell like hot cakes. The age of these companies are ancient, started in mid 90s to mostly early 2000s. The people buying their software don't know what open source is, rather if they see something as free, these are the type of people that think there's something wrong with it. By the articles explanation, they shouldn't exist or they should be fending off attacks from left and right but in reality, the software part of business is actually very small. The fact is, it's a business and the people that make up the customer base. So while it is true that some industries may face fierce competition from "free software" it varies from a spectrum where one end is little to no innovative destruction mostly because of the customers to the other end which this article talks mostly about where ALL software companies are threat. There are still industries where the establishment form coalition and invent complicated "industry standards" to prevent anymore innovation happening. These are business strategies aimed to raise the cost of entry among other tactics that's harmful for the industry but great for the stakeholders.
Having said that, the article isn't without merits, there are some very important truths in there. Like almost all advices, take it with a grain of salt.
WinRAR was a good example. There were free utilities to compress files. Yet, theirs had better interface, features, and performance. The file splitting feature was particularly compelling when dialup, floppies, and CD-R's were common. So, people kept buying them. They've declined quite a bit and we have excellent alternatives such as 7-Zip today. Yet, the product is still selling because they do a job well for a reasonable price.
That's the thing about reading articles like these, they are massive generalizations that apply to big league players. There are still plenty of opportunities to be had for many other players.
That's my take. Funny that it attracts agreement on HN despite counterexamples showing up here regularly. Last one was AeroFS: a private, safer Dropbox for a monthly fee. At the moment, that one use case can (and probably does) have dozens of successful vendors whose sales teams were first to get to customer. Once deployed, they're usually hesitant to switch and have related requirements. That's both increased odds of long-term revenue plus opportunity for profitable, custom work.
That's just one out of a zillion scenarios where potential customer is unlikely to use FOSS or "The Cloud." There's also several attributes outside cost and some business models that almost nobody is pushing hard in software industry like we saw in other industries. Especially retail, which adapted like crazy all over. Anyone spotting these and jumping on them in the right way might make a killing.
Moving forward, assume that middle-men are gone, unless required by legislation or differences in sovereignty (i.e. still effectively legislation). The net result is that it's cheap to build and procure yourself practically almost anything, with the surface of things you can't build/procure yourself increasingly diminishing given the network effect of large swaths of the human race rushing to the Internet to build ever increasingly sophisticated modular components with each trying to outflank each other with a lower price/barrier-to-entry.
If we can 3D print/carve materials we can buy in bulk in a cheap fashion from freely-available software/data, we are likely looking at a fundamentally different economy than the one we are in today.