The Business of Software
Software is a strange business: Software is expected to be free, especially when it’s on the web, yet FAAMG engineers are known for absurdly high salaries. How do these two fit together?
: FAAMG stands for Facebook, Apple, Amazon, Microsoft, and Google/Alphabet: All American tech companies, yet also the 5 most valuable companies in the world.
The Nasdaq Composite is a stock market index which leans towards information technology stocks. In the last 5 years, the Nasdaq has risen from 4.8k to 13.8k, meaning that any investment would have almost tripled, and an annual growth rate of 24%! (calculated on Jun 6 2021). The big tech companies (FAAMG) are the five companies in the American economy with the highest market cap (= the most money invested in them).
WeWork is a startup that provides coworking spaces. This means different companies or individual can rent part of a shared office from them. Despite being bound to the very unsexy and slow-moving economics of real-estate prices, the company branded itself as a fast-growing tech company, and was valued at high multiples of any similar office space companies. How this “tech company” pitch convinced anyone for more than 5 minutes of scrutiny is still beyond me. The most I can come up with is their promise to “optimize work” by tracking work times and interactions, which is just regurgitated “big data” balderdash. Alas, in 2018, the company’s loss was larger than its revenue. When trying to IPO in 2019, the company’s filings revealed widespread mismanagement and misappropriation on behalf of the founder Adam Neumann, and that there was no path to profitability. In 2020, WeWork’s valuation dropped from $47bn to $2.9bn. The WeWork bubble popped.
Further than WeWork, there is great excitement around starting websites, apps, and other digital platforms. Look no further than Gimlet’s The Pitch podcast for a demonstration: Almost all pitched businesses are some sort of app that will somehow use “big data” to somehow make things smarter. VCs (venture capital investors) are pouring trillions into tech startups in the hope of discovering the next Google. (The money behind those investments often comes from unsavory sources such as the Saudi government, but we are getting off-topic.)
The VC model makes sense for software: If you are given a loan to open a restaurant, then if it fails the bank can at least take the building and your inventory. If your social network fails, there is nothing left for the bank to take. Thus software is too risky for traditional loans and more suited for VC money. But the VC model invites unfettered speculation and leaves more risk-averse nations, such as Germany and Japan, behind.
This beats a familiar refrain: The dot-com bubble, which popped in 2000, was driven by excessive speculation on ecommerce-companies and telcos. The belief was that the internet would open up unprecedented new economic opportunity, and whoever was first would win the internet (or something). The poster child for this is Pets.com: The pet supply store took out extremely expensive Super Bowl advertisements featuring their iconic sock puppet, despite their business model consisting of selling pet products at a loss to achieve “growth”, and the hope to make profits once they reached scale.
So are we in a second tech bubble? I’m unsure, but leaning towards yes — while there are more highly innovative tech companies than ever, many phonies are receiving investment, the typical signs of a bubble.
DevOps and the Google way
The B2C paradox: Any investor will tell you that the real money is in B2B (selling to businesses), and that B2C (selling to customers) is much less lucrative, but B2C is what every naive startup founder thinks about because they know those types of businesses from ads and their own daily routine. So: How come most of FAAMG are actually B2C!? Imagine you have a B2B product or service that is 10x better than the competition: Most businesses will actually turn you down, because they don’t want to work with a new company that can disappear, might be locked into a contract, and just prefer the reliability and stability and human relationships of their existing provider. Plus, you’ll have to find a point of contact and do a honkin’ lot of convincing for each new business partner. Some even demand handing over your intellectual property (as AOL did with Google). Contrast that with B2C: Consumers will just switch to the product that is easier to use, and by going online you can instantly reach all 5 billion internet users. So it went with Google: People saw their search was more accurate by a lot, and didn’t have pay-for-placement, so everybody switched. Partnerships with Yahoo and AOL existed, but tough negotiations and a race to the lowest bidder make me question if the effort was worth it.
Pyramids: So, we have a way to market our 10x product, now how do you produce 10x products? Google wanted to get away from the old corporate world of office politics and appearances, and focus on Getting Things Done(TM). One ingredient is hiring the right people, and putting them into the right positions. Google’s interviews don’t let you discuss your rosy resume, but are more akin to exams that test your ability to solve tough abstract algorithmic problems. Using 4 interviewers instead of 1 also ensures that randomness and interviewer personality don’t play a role. A traditional company is built like a pyramid: You can only advance by becoming more senior, and as a bottom layer of the pyramid you will be poorly paid, no matter your output. Google’s innovation is promoting bright people quickly, even if they are young.
DevOps: Another ingredient to rapid growth and time-to-market is automating all parts of the software development lifecycle, with testing and sysadminning being the most obvious targets: This is DevOps. In fact, Google invented DevOps before DevOps was a thing (2009), so they have a separate term for it — Site Reliability Engineering (SRE) (2003). Kubernetes is a Google project. I read a criticism of DevOps, which bemoaned the fact it always gets compared to the waterfall model, “an entirely fictional model which no company uses”. I’ve worked in companies that used the waterfall model, and that was in 2020. DevOps is far from the norm.
Free: Google has massive warchests from their advertising business, so they produce all kinds of internet services which are not only much better than the competition, but also free. Think Mail, Maps, News, Docs,1 and the classic Search. Google makes things free to make the internet more attractive, leading to more people online for a longer time, which in turn benefits Google as they own over a third of all digital advertising. This drives SMEs who cannot offer their software for free out of business. In 2000, Ev Williams published a now-removed article called “The End Of Free”, where he prophesied that free software was a figment of the dot-com bubble, and that we are reverting to a paid model. He later became co-founder of the free(mium) sites Twitter and Medium (it goes full circle!).
My hot take: Google is a true innovator able able to do things so differently, and grow both stock price and revenue 30% annually for now 13 years straight. WeWork and others are impostors which promise the same amazing growth Google had but will not succeed.
(For more insights on how Google does things differently, I can recommend the book I’m Feeling Lucky by Douglas Edwards. I recommend skipping most of the memoire-y stuff however.)
Software is increasingly offered in a SaaS (software as a service) model. This means you don’t purchase the software as a product and install it on your private PC or company server, but you are paying a subscription for the software developers to also host it for you. As an example, Microsoft has pivoted from selling new versions of their Office suite every few years to a yearly “Office 365” subscription (although you can still get a license via one-off payment).
This has obvious advantages. Internal techs often don’t do a great job of hosting, so having the vendor do it usually provides a smoother and more updated experience (this goes hand-in-hand with DevOps).
It is a well-known psychological trick that many small payments are much easier to stomach than one large one. Go ahead and sum up how much you’ve paid in total for internet services, I’ll wait. Another trick is loss aversion: If you cancel the subscription, you lose something, may it be access or even just a special badge, and are treated like any non-paying customer, no matter how much you “invested” in the past. This feels incredibly bad, even if you are not using the service anymore. Are tech company profits just made up of subscriptions we can’t let go of, or even worse, completely forgotten subscriptions that are still silently charging our account?
Games as a service: Particularly evil are games as a service. While the business model does make sense for multiplayer experiences that run on paid servers and are continually evolving, a lot of games do not implement this as a flat monthly fee, but try to extract as much money from the player as possible. This often follows the freemium model: The base game is offered for free, and some features, items or (oh no) lootboxes are paid. Most people think to themselves (rightly) that they would never pay into a stupid free game, so the tactic they developed is devious: Completely consume the player, have them put 100s or frequently 1,000s of hours into your game, at which point they can justify the investment with how much time they spend on it after all. Those hours are often filled with incessant nagging to drive a purchase, so they are not designed to be fun but only habit-forming. While a product game is like a nice vacation, a service game is like legcuffs, or a smoking habit.
Twitch.tv: With GaaS, studies show that most revenue actually comes from a small sliver of the playerbase, who will spend 1,000s of dollars on it, the so-called “whales” (a term from poker — many GaaS seem to imitate casinos). I’ve seen a similar effect on Twitch: Users in chat can be generous to the streamer and other users by “gifting a sub”: This means spending $5 to allow a random or chosen user to become a subscriber of that streamer which comes with a few nice perks, plus the streamer gets around half of the money. Twitch publicly shows gift leaderboards, which encourages people to compete by gifting subs, but also shows me that the streamer’s income is driven in large part by a handful of “whales”. Whenever someone gifts a lot of subs, chat automatically assumes that it must just be a bored rich person, but more often than not it is a normal person spending unreasonable amounts of money, just like in GaaS. I know a guy who spends around CAD 600 a month on Twitch, which is almost his entire income. And even I have spent more than I’d like to admit. How many internet companies are profitable on the backs of addicting and financially ruining a few individuals, while wasting everyone else’s time?
So, where do I see myself in all this? I for one am tired of working for mediocre German companies, with half-competent colleagues, on banal internal projects. Ideally, I’d avoid working on something that actively addicts people, destroys their social ties, and ruins their life, as so much tech seems to be in the business of. Ideally, it would also address climate change (ironically enough, only refraining using resources and doing business would truly achieve this: computers do not appear out of thin air, they need to be produced).
The problem I see in Germany is that there is very little expertise in old economy companies. Expertise tends to be in consulting companies (that is their selling point after all), but consulting is bounded by hourly rates, so unlike with a product or service company, doing an amazing job does not pay off. There isn’t much in the way of startup and new economy companies either (especially if you are not willing to move to Berlin, which I am very much not). The only ways for advancement are climbing up the org chart (which gets increasingly harder as it gets increasingly narrower, like a pyramid) or starting and growing your own pyramid (where you’re guaranteed a top spot).
So, starting my own company it is then, right? I do care a lot about how things are done, I’m very technical, and I received excellent grades in uni, so I definitely have the skillset. The two main issues I see is (1) I have laughably low self-esteem and (2) I don’t have a network of other people I could collaborate with. If I were to break into any industry it would be extremely valuable to have an experienced partner, but that kind of person would be unlikely to join a startup and listen to a 23-year-old. This only leaves pure software/internet plays, but that is an extremely competitive field, and still requires competent partners. I guess sometimes you have to let dreams be dreams.