Hacker News new | comments | ask | show | jobs | submit login
The Cost of Developers (stratechery.com)
397 points by djug 8 months ago | hide | past | web | favorite | 227 comments

Great analysis, although I don't completely agree with this statement: "GitHub, a company that, having raised $350 million in venture capital, was not going to make it as an independent entity."

If it is referring to the fact that, in the crazy VC spiral of the startups world, once you have received such a big investment, a sale to a big corp is the only choice, then I sadly agree. But GitHub could have been an independent entity with a sustainable business, if they had not fallen into VC's startups trap. I have to agree with DHH here: https://twitter.com/dhh/status/1003611913924894720. We need more independent software companies instead of only having 5 big players.

Correct, github sold out to investors, which then owned them, which then sold them out to get an exit when it turned out the company was in no shape for an IPO which is the only reasonable other exit if you are looking for a 10x ROI on a company that burned through hundreds of millions. Investors were in this for a huge exit and they just got it. In fairness, MS bought a valuable social network of essentially the entire OSS and developer community.

As an independent entity, Github would have had to evolve in a very different way and gone more aggressively after making money, fixing their cost structure, like most of their competitors like Atlassian, Gitlab, and other companies selling developer tools as a SAAS service. So, even though these are comparable companies with comparable offerings, the value proposition is very different. There is no way in hell Gitlab is worth anywhere near what MS just paid for Github.

Being based in silicon valley means Github paid a premium for being there and burned through loads of cash paying for expensive developers, fancy office space, etc. Much cheaper if you are based elsewhere, have less emphasis on wanting to have every OSS project on your platform and more emphasis on selling tools to corporations. This is in a nutshell the strategy for Atlassian and Gitlab. Both also offer freemium layers for OSS but mostly their deal is upselling to their paid offerings. Growth that way is much slower. Though, I would say, Gitlab is now pretty well positioned to succeed where Github struggled.

>, github sold out to investors,

>Investors were in this for a huge exit

I think these excerpts from your comment and also DHH's "VCs need their pound of flesh" are not helpful for readers on how to analyze the situation. They (maybe unintentionally) taint the discussion.

Github is an entity owned by human beings. The founders included Chris Wanstrath, Tom Preston-Werner, PJ Hyett, and Scott Chacon. Therefore, emphasizing that "investors wanted a big exit" is (inadvertently) omitting that those 4 founders may have also wanted a big exit as well. The DHH quote also misdirects people into thinking the VCs are the bad guys. Instead, we have to remember that the 4 founders have to sign off on the documents at closing to get the $350 million. They had to do presentations to convince investors to give them $350 million.

We have to include the founders' thinking about Github's future and not just outsource our frustration to those "evil VCs". VCs cannot give money to a startup if the founders don't want it. (E.g. Craig Newmark refuses VC money for craigslist.org.)

You are extrapolating an opinion that I don't have. I don't consider the founders, investors, or MS evil. The only mildly controversial thing here is that a big corporation like Microsoft with a long history of hostility towards anything OSS ended up buying the goto gathering place for the OSS community. I don't think that ever was a dream scenario for any of the early founders or early adopters of that company. It will be interesting to see if e.g. Linus Torvalds migrates his linux repository elsewhere.

But obviously, Github was never a charity. It is owned by a group of investors that invested hundreds of millions. An exit in the form of an acquisition or IPO are the only two reasonable happy endings for such a company. This was like this from the day the founders decided to sell ownership to investors. That's what it means to play that game. Usually a first round leads to follow up rounds. Which leads to revenue and ultimately to an IPO. That's the scenario you pitch to investors. In the end it is about shareholder value.

So, normally an acquisition is considered the less ideal outcome of those two. It implies some sort of failure to deliver. Github running without a CEO for a while, reported cash flow issues, and then getting acquired by MS sort of supports that kind of story. I'm assuming the founders did indeed well for themselves. I'm wondering if the same applies to their employee stock option programs since typically those are a bit lower in the pecking order. I don't really know enough about the ownership structure to say anything here but I imagine for most employees ending up working for Microsoft might not have been the dream outcome. And as for most Github users, this probably was a bit of an surprise/shock.

In my case, I'm a paying customer and I see no reason to change that. Love the product so far and MS seem to have the right intentions and strategy to keep it that way.

> I don't consider the founders, investors, or MS evil.

Fair point if I overstated your perspective on VCs.

I don't know if English is your primary language but if you didn't know, the phrase "sold out" is virtually always used in a derogatory manner. When one writes "<X> sold out to <Y>", the <Y> is the proverbial bad guy. It's also parsed as an insult to both <X> and <Y>. E.g. "It's a shame that guitarist sold out to BadMusic Record Company."

It seemed like "sold out to investors" emotionally taints the VCs as bad actors and poisoned the discussion. (Yes, some VCs are bad and unethical.)

In any case, people should think of VCs as one type of financing option. (Bootstrap is another, but that has its own risks and downsides. Bank loans are another finance option but most startup founders have no collateral.) If you don't want to pressure your company into a big exit, do not talk to VCs.

>So, normally an acquisition is considered the less ideal outcome of those two. It implies some sort of failure to deliver.

Maybe. One of the VCs I saw mentioned that acquisitions are actually the more common exits for B2B companies. Github (enterprise) was more B2B than B2C. If this is the consensus view, both a16z and Sequoia may have anticipated acquisition as the most likely (and successful) outcome at the time they invested.

Well, the founders certainly are (as near as I can tell) letting MSFT and their VCs take heat for a decision the founders alone made.

Taking $350m in VC means you promise to build a stupidly profitable company or sell it to a high bidder. That's the deal, and it's completely understood by anyone in a position to raise nine figures of VC.

I'm not native indeed but fluent enough to deal with the subtleties. I like to use strong language when making points in public fora. It's a style that is admittedly a bit over dramatic but tends to provoke people a bit more into engaging.

From my dealings with VCs my understanding is that they are mainly incentivized by investing and less by the success of those investments because the timelines are half a decade or more and they reap their rewards long before that happens. Basically the job of a VC is to spend money in some fund. It's a very competitive market to be in. Acquisitions are the tool of choice to make sure under-performing investments are not a total loss. A lot of acquisitions and acquihires are effectively investors consolidating their losses. It's common for them to have stakes in both the acquiring and acquired company. In this case, that's probably not what happened of course. But still, the investors are probably quite happy with this outcome.

From the standpoint of an end user (which is all of the people complaining about Microsoft acquiring Github) VCs and their money are the bad guys, because the dynamics of their investment is what causes the acquisitions which change the service in ways that I (the end user) do not like.

Not saying they're right or wrong (I think at the very least it's a limited point of view) but that's the argument being made.

>VCs and their money are the bad guys, because the dynamics of their [the VCs] investment

You're repeating the same "outsourcing of blame" to VCs that I was trying to short circuit.

VCs as an abstraction are a convenient target but that doesn't mean they are the correct target.

The founders of Github such as Chris Wanstrath, Tom Preston-Werner, et al have agency and autonomy. Why are they specifically not included in our analysis?

In other words, why aren't we saying this:

>CW & TPW's desire for VC money are the bad guys, because the dynamics of their request for VC investment

It seems wrong that analysis, blame, and frustration seems to gravitate towards VCs and terminates only at the VCs -- instead of the founders who willingly asked the VCs for the money.

Yes, DHH is against VC money. But maybe CW and TPW don't think about Github exactly the same way as DHH thinks about Basecamp.

VCs are a convenient abstract bogeyman. Yes, VCs do tend to encourage a grow big or go home mentality but it is, as you say, the founders who wanted that sort of investment.

It's also the case that, as soon as someone here suggests that a startup doesn't need to operate in that manner, someone--as certainly as the sun will rise in the east--will post Paul Graham's startup definition and argue that you're not a real startup without a growth-first mentality.

You're super-close. Money is the "bad guy". It might not appear to have agency or autonomy, but that is because humans are not good at perceiving intangible power structures.

In other words, why aren't we saying this:

>Money is the bad guy, because its presence, uneven distribution, cult of worshipers, fungibility, and inherent ties to social power structures are unhealthy influences on the founders of Github

I feel like I should post Francisco D'Anconia's Money speech here....

Money by itself isn't the "bad guy"; its a store of value. Like Technology it can be used for good or bad. We have structured our society to reward greed with unbelievable wealth. Like how can someone spend billions of dollars of their wealth? What is the need for someone to own all that wealth? And yet we all take that as given and aspire to it.

Beyond money, it's the logic of commoditization that is the bad guy.

I can't think of anybody that would be better to acquire GitHub, if it had to be sold off.

At least Microsoft is heavily invested in developer tools, and making them better all the time.

Users love the VC's money. Uber handed out multiple billions of dollars subsidizing both riders and drivers from these massive investments, but even smaller investments often mean the same things. From photo sharing to local delivery most of these services are beloved in large part because they don't need to be directly profitable.

...yet. "Directly or indirectly profitable" eventually happens.

Okay, but anyone making that argument should also consider that the existence of VCs and their money may well have been part of the reasoning that led the founders to create the service in the first place. Look at the almost total absence of such useful or interesting services created in places that don't have VCs.

The term, "github sold out to investors," adequately includes the founders as part of the equation to me. "Sold out," describes all that you mentioned about courting investors and whatnot.

At where I work we use Github Enterprise, which costs way more than Gitlab Enterprise, is not scalable, and barely cares about enterprise customers like ourselves. Gitlab on the other hand is much cheaper, is highly scalable, runs in AWS, and is responsive to requests. Ultimately to make it as a business you need to provide what the customer wants and is willing to pay. Github never did.

We switched from GitHub Enterprise to BitBucket Server several years ago and are pretty (80%) happy with the decision. Saved an absolute truck load of money.

The only issue with BitBucket Server is that it (inexplicably) does not have a commit compare view.


This is such a massive oversight I have to conclude Atlassian themselves don’t use this product.

It annoys us but not enough to pay GH.

We don’t like GitLab as it was slow when we tested it and the UI wasn’t designed as nicely as the other two (tbh it looks like it was designed by programmers). Something about GitLab feels off and I don’t really know what it is.

If MS changes the cost structure of GH Enterprise we’d probably switch back (haven’t looked into how well it integrates with Jira etc recently).

Also, their Sourcetree offering for Windows is quite nice, but the OS X version has big problem with git-flow.

You can tell it is two very different teams working on it.

Atlassian is now like the Cisco of late 90s an early 2000s, an amalgamation of disparate companies' technologies under the umbrella brand.

There are many things I find frustrating about BitBucket Server (it’s super slow, for one, it’s insistence on clobbering commit strings during merges is the biggest), but I’ve never wanted a commit compare view; I don’t even know what that is.

Care to explain what you would use it for?

Just seeing all the commits/changes between two commits.

So with GitHub Enterprise when you push multiple commits, the webhook sends a “compare” URL which is a handy look at everything that got pushed.

It’s not possible to do the same thing with BitBucket Server.

You can compare two tags and get the same view. Or two branches. But not two commits.

I'd be interested to hear about the slow behavior you're seeing. There are some tips and tricks for improving that.

Generally speaking, I have found Atlassian’s tools don’t scale well to megacorp usage. When there are thousands to tens of thousands of users connecting at peak times, stuff just gets extremely slow. Noticeably slower than when we all put up our own team-level Phabricator instances, although I’m obviously not adjusting for Phabricator downtime when inevitably the unofficial team sysadmin went on vacation and the server fell over.

I’m sure people involved in infrastructure are working on making this better, but I don’t have any visibility into that layer.

From a speed standpoint, I'll say the obvious thing, "it's architecture dependent".

For thousands to tens of thousands of users, most folks need data center with 1+n nodes (w/ larger diminishing returns at 8 nodes). There's a few tips and tricks with that, like pointing CI/CD tools to a smart mirror instead of the primary instance and only mirroring specific build projects. Putting smart mirrors in the same physical location as a primary instance could also help reduce the impact of CPU-intensive requests like git fetch or clone.

Another common issue is apps/plugins. Bitbucket 5.9 added a way to get app diagnostics so infra folks can see if there's a long-running process or other weirdness. Vanilla Bitbucket needs about a 1gb heap (java) and with apps, you want around 4gb (give or take).

If you want a longer read to share, this still has some good info despite being 2~ years old: https://developer.atlassian.com/blog/2016/12/how-we-built-bi...

> The only issue with BitBucket Server is that it (inexplicably) does not have a commit compare view.

Github and Github Enterprise also fail to provide this view. They do provide a compare view for the triple-dot notation (git diff <commit1>...<commit2>) which actually compares the merge-base of commit1 and commit2 with commit2. It's not possible to do the equivalent of git diff commit1..commit2 in the web interface.

> If MS changes the cost structure of GH Enterprise we’d probably switch back (haven’t looked into how well it integrates with Jira etc recently).

Given that they acquired GH mostly for the Open Source parts, I wouldn't bet on that :/. Enterprise was Github's way of making money and that incentive is kinda reduced, if not eliminated entirely.

Atlassian employee that works with/on Bitbucket here.

We're working on improving that experience. I'd keep a lookout on the next few releases.

I was on a team evaluating Github and Gitlab for a large corporation (>10000 seats). Github was multiples more expensive than Gitlab, and unwilling to even come close on a counter offer.

I am a fan of Github (and can appreciate their reluctance to deal with Enterprise headaches), but Gitlab was much more engaged and committed to win the contract. Github didn't seem to care whether they got it or not.

There are quite a few things about gitlab that smell of amateur hour with the way it's put together, but on the whole it's a really good product and I like that enough to make it usable is free open source, and the CI stuff is much nicer than any other I have used (although the way we've put it together, we've hit its limitations a bit).

You accuse Gitlab of "quite a few [problematic] things" but do not put forward even a single one of those.

I was on mobile and on my way to bed :P. Lots of page refresh related problems, and likes to exhaust memory from time to time are the two standouts.

Have I missed something here? Why are you with Github and not Gitlab?

Generally speaking, the set of people who live with the consequences of a decision is not the same set of people who made said decision.

The commenter may not be in a position to make the decision to switch. It's also possible that changing would incur high switching costs - although I'm not sure what they'd be.

We have both, however some teams predate Gitlab option and corporate inertia, politics and a lack of time keep those teams from switching. In a huge company changing anything can be a major pain in the ass.

Corporate inertia?

Gitlab also took VC money and have an exit as their only option. Doubt they'll get as far as GitHub did though, and the option to just grow slowly into a solid independent business is off the table. Probably an acquihire with small return to investors.

Gitlab is really open about building to a 2020 IPO

"We want to IPO in 2020, specifically on Wednesday November 18. 2020 is five years after the first people got stock options with 4 years of vesting. To IPO we need more than $100m in revenue. To achieve that we want to double Incremental Annual Contract Value (IACV) every year. We focus on an incremental number instead of growth of our Annual Recurring Revenue (ARR) because ARR growth is misleading. So far we achieved the goal of doubling IACV in 2013, 2014, 2015, 2016, and 2017."

Great find! Here's the source link: https://about.gitlab.com/strategy/#sequence-

A big difference being Gitlab is open source and can be self-hosted, so you can still run Gitlab on your own terms even after (insert faceless corporate entity here) buys it.

If you care about such things, anyway.

> There is no way in hell Gitlab is worth anywhere near what MS just paid for Github.

On the other hand, Atlassian is worth twice what MS just paid for Github.

> Gitlab is now pretty well positioned to succeed where Github struggled.

Gitlab is also VC funded. In fact it seems like you should be _happy_ that Github is no longer owned by the VCs.

I agree with one caveat, I don't think it's the amount of money. I think it's valuations, and methods of funding/valuing companies.

Zuck still owns 30% of fb, after (I assume) cashing out some shares. Early execs/investors would probably own 75% or more of the company if you exclude shares sold/cashed out (as opposed to dilution).

This is because that FB never had to raise serious^ money. Put another way, it does not cost money to make a FB. This makes sense, FB is a "regular" website/app and those are cheap to make.

You could say the best site wins, the most competent team. You could say it's a lottery. Either way, one site gets to be the social media site. That's valuable. Money is not required to get there.

Compare this to TSLA, on the other extreme. Musk sold all the shares immediately. Then he borrowed as much as he could. This is because it takes money to build a TSLA. You need factories, parts, a supply chain... expensive.

This means capital is being allocated very inefficienttly. FB could be the same fb (to users) @ a tiny fraction (maybe as low as 1%) of its market value. Twitter too. The money that is going into FB (and into shareholder pockets) does not enable FB to exist, help it do more things or benefit consumers/the economy.

If FB's valuations was much lower than it is, FB could have done the exact same things that it did. If Tesla's valuation was much lower, they would have had less capital to invest and they'd be making fewer cars. A dollar that goes into FB makes FB shareholders richer, with no effect on anything else. A dollar invested into Tesla makes cars.

Classical economics isn't supposed to work like this. The market is supposed to allocate capital where it is needed, at least in broad strokes.

I think that we're dealing with 3 distinct things.

(1)the monopoly-like nature of the digital communications economy is having hugely distortive effects. (2) Centralisation of capital into giant pools means money is flowing into giant "investment vehicles." Big investors = big funds = big companies. (3) we're in the middle of a capital bubble. Returns on doing business (selling stuff to people for money) are not as big as returns on investing (selling stuff to investors for shares and other, non-money money). 1 unit of "capital" is worth more today than yesterday. Ie inflation. This is squeezing out the "real" economic activity.

^scaled to market cap.

>This is because that FB never had to raise serious^ money. Put another way, it does not cost money to make a FB.

Even considering your footnote about market cap, it was actually very expensive to run Facebook.

Facebook started in 2004 and didn't turn a profit until 2009. For more than 5 years, they were burning investors' cash on infrastructure. They quickly burned through the 2005 $12.5 million investment from Accel.

Probably the #1 stress on infrastructure costs was photos. In 2005, Facebook added a feature for people to upload unlimited photos. They soon had more photos than Flickr. As we've seen from other flavor-of-the-month photo bucket websites that crashed & burned by getting overloaded traffic from reddit users, hosting billions of images is not cheap.

Did all of that go into running FB? No cash outs for early employees/investors?

In any case, that is still a small portion of their eventual market cap (2%). Also, they had it. It was cheap money. Why not spend it. If they didn't have it, they may have had to create a revenue stream or control costs a little. Running for years before considering revenue or cost control is a luxury.

When you're cashed up, you get spendy.

But, point taken. It's all matter of degrees. It did cost something to make FB. It just wasn't much compared to most companies its size.

Zuck has a large stake because he got his seed funding from his parents and could negotiate their series A round from a position of strength.

Musk started with $100m of his own money and a 1-in-a-billion reputation. He had leverage of that kind too, I would say far more than zuck.

The real leverage was not needing money after that A round.

Musk didn't start with $100 million. He started his second round of start ups with the money made from the first ones. And he didn't have $100 million for the first round of start ups he founded.

In fact Musk has pretty much suffered through all the regular pain founders suffer through.

I remember watching somewhere on Youtube, during his first start up, they would code on a computer by the night and host the application for their users during the day on the same computer.

The social media industry is still nascent, it takes a lot more time than Facebook and Myspace have had to grow an entire market segment. Eventually we'll see a plurality of options and the ability to move between them with relative ease. But it's already easy to ignore Facebook, I didn't think that would happen for at least another decade.

In fact, the only reason I'm still on it is because it maintains a network of former friends and contacts that no other service has been able to offer. Make no mistake, Facebook is struggling for relevance in an age where the next generation has already moved on to Snapchat and friends.

I don't think any of them are going to emerge dominant, but if Facebook is smart, it'll use it's remaining heft to build a true platform instead of trying to keep going as an aggregator.

Relevance can mean different things. Maybe FB is not "relevant" in a specific cultural sense, the avante guard users don't care about it anymore. Maybe that says something about the future. But...

It is the only relevant social network if you want to sell concert tickets, soap or win an election. For hundreds of millions of people FB is the internet, because of "free data" plans in 3rd world countries. It is the only relevant social network in terms of revenue, market cap, M&A and a lot of other things that are unrelated to cultural relevance.

Sure. For now.

I agree. But, there is this narrative in the VC world that the tech game is different than ordinary economy. In tech, you have to create a monopoly due to network effects. Brian Arthur has been the first to formulate that [1], and it has become SV folklore ever since.

If you believe in that worldview, there is no space for a market with several players. The only goal must be to become the number 1 as fast as possible. See also “From zero to one” by Thiel. Needless to say that this is a depressing worldview, but either they right and then it won’t help to whine about it. Or they ain’t right, but they still gonna operate like it is true. Both sad.

[1] http://a16z.com/2018/05/16/network-effects-positive-feedback...

Yes but given the opportunity, between retiring early with hundred of millions and keeping working for a few millions, most people will choose the former.

> the App Store dramatically lowered the barriers to entry for developers

Is there any platform with higher barriers to entry than the App Store? iOS development requires a proprietary toolchain only available on Apple hardware, programming languages hardly used elsewhere, not to mention the $99/year developer fee and the litany of vague and arbitrarily enforced rules and content guidelines for App Store release.

stratechery wrote: "the App Store dramatically lowered the barriers to entry for developers"

To which you replied: "Is there any platform with higher barriers to entry than the App Store? [...] , not to mention the $99/year developer fee"

His baseline for "lowered barriers" was retail channel distribution.[1] Think of shrinkwrapped packaged software like Quicken on the shelves of CompUSA, OfficeDepot, Costco, and boxes of videogames at GameStop.

Yes, $99 is a barrier but it's much less than the millions of dollars in capital required to get retail stores to carry your boxed software.

[1] if you click on the url link behind "lowered the barriers", you'll be directed to this page illustrating retail distribution: https://stratechery.com/2013/friction/

Interesting. In hindsight, the retail store distribution model of software lasted far longer (and still does) than it ever should have.

I think you are underestimating how much easier high speed internet enabled downloading of software over buying a CD

Yeah, I very clearly remember a "the sky is falling" sentiment when Steam launched a download only version of Half-Life 2.

Or take Netflix, very few thought that would be the streaming juggernaut that it is into today. Alas poor Blockbuster, we knew you well.

The telcos were taking 70% of the gross before the Apple store came along. A big reason people flocked to iOS was the hope of finally turning a profit, because Apple more than doubled the developer’s take. That was a huge barrier.

What % is Apple taking now? I know they lowered it a bit but I don’t recall what the new numbers ended up at.

Before the App Store selling mobile apps required negotiating with every carrier individually, and they priced based on your perceived ability to pay. We had clients who were interested until they were quoted around $50k per carrier just for the privilege of being listed!

The old Palm smartphones let you run whatever you wanted on them.

Yes but relatively few people had them and the carriers were trending in the direction of taking away flexibility. If Palm had hung around longer, I'd bet that one of them would have tried the locked down store there as well.

Compared to previous mobile phone development? Yes, absolutely.

Compared to getting into PC retail sales? Yes, absolutely.

Compared to web development? No barriers at all, but hard to earn money other than by ads.

Compared to PC shareware? Well, that did have a lower barrier to entry and did make lots of money for id Software. This might genuinely have been the lowest barrier to entry platform that it was possible to earn money on.

I'd never made an app, never written C#, and i got an app on the appstore within a few weeks and being downloaded around the world soon after. And that was 7 years ago before they made dealing with all the certificates not insanely annoying. If that's not a low barrier to entry i don't know what is.

In a way the barrier has increased recently with a lot of new requirements for apps (designing for the notch etc) and appstore listings and worrying about privacy policies etc.

I think 7 years ago it was a matter of luck.

Does your app look good? Yes? Ok-

Did your app overlap with Apple? No? Ok-

Release your app! And thanks for paying apple 2500 dollars in various unavoidable payments.

Hoping React Native Javascript saves me from apple.

Video game console development historically has often required thousands of dollars in special equipment and sometimes even a special relationship with the console manufacturer.

Good point. So if our point of comparison is video game consoles, the barriers to entry have indeed been lowered. But is that the right point of comparison?

Before the App Store, you had to have relationships with each carrier to get on thier app stores or you could write independent apps but no one would find you. They also took a 70% cut - not a 30% cut.

It's worth bearing this in mind. Almost everyone expected the App Store fee to be 50%. Opening at 30% was considered stunningly low, to the point where it was almost considered unfair to other platform owners because it was felt Apple was subsidizing the system, taking profitability out of the App store business.

I remember reading in Steve Jobs biography book that he had the rule of making a 30% of sales in deals with third parties. Sometimes this was not remotely possible so in those cases the numbers had to be massaged so that he could see his 30% and approve the deal.

The barrier was the _distribution_. Before the app store, you'd have to make a website of some sort to sell your app. And then you'd have to advertise your app, and so on. The app store does all that for you, so that you just build your app and that's it.

"Is there any platform with higher barriers to entry than the App Store? "

Just about everything that came before. Getting an app on a phone before meant you had to go through the carrier's store, where they had a 70/30 split, with you getting 30%. There's also selling your software in a box on a store shelf, where the distributor takes 50% off the top.

Barriers are things that prevent you from doing stuff. To start a bank, you need billions, a reputation and friends in the industry and central bank approval. Even a whip smart billionaire would struggle.

Getting into the app store requires £1k of hardware, enough programming chops to learn a new language and the sort of bureaucratic hurdles you might encounter signing up to a cable tv contract. If you don't have the chops you can hire them.

Apple do control the whole system via a tight bottleneck, but barriers aren't high. The web is lower still, but that's about the lowest barrier to entry for anything.

I think starting a bank involves high hundreds of thousands, or millions - but I'm in the US - starting a bank is actually pretty easy here compared to other countries.

LOL. I remember giving Ingram MicroD 50% of sales just to get stocked, then spending tens of thousands on advertisements, and on store sales promotions just to get them to order our software from Ingram MicroD. That after spending tens of thousands on printing manuals, boxes and duping disks.

And you want to talk about vague and arbitrary rules, just try to figure out why the stores that said they’d promote your software buried it in the back and returned it after cashing your checks.

My favorite was trying to get paid by Ingram MicroD. “It’s been 90 days, you agreed to pay within 45”.

“Oh sorry it’s friday, we only cut checks on Thursdays”

“Oh, you didn’t get your check? I’m sorry, I’d like to help but it’s thursday, we only cut checks on Tuesdays ..... oh I don’t know why she told you that, we never cut checks on Thursdays”

“The good news is I went ahead and cut your check, even though it’s only Wednesday. The bad news is our CEO is out of the office, so it isn’t signed yet ..... oh, I’m nit sure when he’ll be back, it’s a family vacation.”

Video game consoles require approval before they will even send the SDKs.

MS paid $7.5 Billion for Github, along with its 28 Million users - it's in the first paragraph of the announcement (1)

That's around $268 per Github user.

Now MS could have probably made a site comparable to Github for 0.1% of that cost - e.g. revive codeplex, pump a few million into the design, backend, git support and user interactions.

But it wouldn't have 28M users. That's what they're paying for. That's the hard part.

1) https://news.microsoft.com/2018/06/04/microsoft-to-acquire-g...

> That's around $268 per Github user.

That seems like a decent deal for MS... Assume on average a user is an active engineer for 5 years (of course this is back-of-the-envelope and not even taking into account the growth of the field, and new engineers are likely to use Github for it's network effects).

Is the value of owning the central platform of an individual engineer worth more than $50 / year? Even if all you did was run basic ads I'd say yes - although I'd expect more services including edging in on Atlassian's turf as an upsell opportunity for MS.

All this, not even counting future value... If Slashdot is still around and somewhat valuable, Github will probably also still be around and somewhat valuable at least a decade from now.

> That seems like a decent deal for MS..

Indeed, as cost of acquisition of an active developer, it doesn't seem either absurdly high or absurdly low.

They did, it was called Visual Studio Team Services.


And it's awful. Not even comparable to Github and GitLab

I second this. I tried to use it when it was launched because it purported support for a build farm, but it was too terrible to figure out how to use it and I gave up after a couple of hours of trying to figure out how to make it do ANYTHING useful.

You should give it back a try, they have been improved quite a lot since all these years.

All these years being the 2 years since it was released ?

There are times when I really like the thoughtful analysis on Stratechery and there are other times when I think it's a complete pile of detached-from-reality horsesh!t. Today is of the 2nd kind.

What it also has is a potentially fatal weakness: no platform with user-based leverage.

Sorry, what? Windows still has > 90% market share on desktop. If that isn't "user-based leverage" I don't know what is.

This, by the way, is precisely why Microsoft is the best possible acquirer for GitHub, a company that, having raised $350 million in venture capital, was not going to make it as an independent entity.

Seriously? GitHub could have easily gone out for more VC. They could have IPOed. "Not going to make it as an independent entity" is so far detached from reality, what was he thinking when writing that?

Github would have had trouble getting more VC investment. According to Bloomberg, there are ~21M software developers in the world, and Github has ~24M user accounts. There's little userbase growth possible there; they would have had to start increasing revenue per user rapidly to be interesting for another VC round.

Equally, on their $200M revenue (or whatever it was), they would struggle to get a multiple that would be interesting for their VC backers. Their market cap at IPO wouldn't have been much more than $1B (absent some growth story).

It's a great problem to have, to be sure, but the MS purchase is a great deal for Github investors.

They could have easily become a formidable competitor to Atlassian if they extended their offering beyond just code hosting. Competing with Atlassian is actually not that hard considering how bad Atlassian is at times (I'm a customer myself). Atlassian today has a market cap of 15B.

There's a lot of growth opportunities if you have >20m users.

I used to work for an Atlassian competitor and the feeling in the company was a sense of intimidation.

Now that I’m a customer, it’s really ridiculous how badly their stuff is integrated. I mean, when I find myself wishing for the good old days of using Trac pre-1.0 because at least it fucking works and I don’t have to log in four times while working a ticket through code review and into a deployment. I mean Jesus Christ Atlassian, you're on version 6 and you can’t make SSO painless, and every time I type in a ticket number into an input field behaves differently.

Why hasn’t a competitor eaten their lunch yet?

Agree. There are some easy project management wins for Github. It’s flabbergasting that they’ve ignored them. Issue complexity and time estimates. Burn downs. Github is only part way there, why not get all the way?

Sometimes it's worth it to stick to what you're good at.

If Github burnt a ton of cash to try and compete with Jira, ended up making a product that didn't attract enough people over at the same time as ignoring problems that users were already grumbling about, this could have led to them being in a worse state and easier for MS to pick up on the cheap.

If it cost a ton of cash, that would be ridiculous. It is adding some fields and charts. Do you think Zenhub has or had a ton of cash?

If it was so easy to make a decent project management tool, there would be one...

There are many. We built one in Google Sheets. Zenhub layers over Github. We love it too. But I am sure it didn't require tens of millions of dollars to accomplish.

I've 4 Github accounts, 3 for past companies that I've worked at and 1 that's personal. I wonder how many unique accounts are on there. I've interviewed plenty of devs who say they don't have a github account.

According to Tim Cook yesterday, there are 20M Apple Developers. Someone is not telling it like it is :)

If that is based on number of Xcode downloads then it will over-count, you need to install it to use Homebrew.

I use Xcode as the command line tool chain to avoid needing brew.

Only recently did I discover you can download the CLI without installing XCode first. Saves about an hour off of setting up a new developer.

is this a joke? Investors were begging to get in Github it's whole history, no surprise it sold for 7.5 billion

> Sorry, what? Windows still has > 90% market share on desktop. If that isn't "user-based leverage" I don't know what is.

At first I thought the same way, but then I started wondering if those users are really meaningful users. Isn't most of the new developer products are either mobile apps or SaaS solutions, leaving aside monsters like adobe? If so, does it really matters that so many users are using windows if they consume apps mainly with their (probably, chrome) browser?

> If so, does it really matters that so many users are using windows if they consume apps mainly with their (probably, chrome) browser?

Windows is monetizing Telemetry data, which does have meaning and value, even for people that never leave chrome.

As far as we, the general public, know for certain, telemetry is used to gather usability and preferences data, and bugs.

You can avoid using their software if you are concerned about your privacy, that's totally legitimate, You can also warn others about it, if you wish so and believe that it is an important cause. I'll walk most of that walk with you. But I believe you have no justification for your accusation.


This suggests it is used for advertisements.

[citation needed]

Sorry, what? Windows still has > 90% market share on desktop. If that isn't "user-based leverage" I don't know what is.

In the grand scheme of things. No one cares about desktop development. Walk into a VCs office and tell them that you have the next great idea for a Windows app and watch how fast you get laughed out of the office.

How many developers are going to flock to writing Windows apps?

As much as I hate it, most of the popular desktop apps are written using cross platform toolkits like Electron. The only companies making serious money on the desktop are Microsoft and Adobe.

Games don't count as "desktop apps". They are mostly using cross platform game engines and not dependent on being part of the Windows desktop.

As far as IPOing, if they couldn't make money as a public company, they would have been an acquisition target either way.

>Walk into a VCs office and tell them that you have the next great idea for a Windows app and watch how fast you get laughed out of the office.

In most of the world the "goal" isn't to go raise VC money. It's to develop a product and earn a living.

Maybe people building Windows apps for the massive market are doing okay without the desire to raise billions and take over the world?


This is an excellent overview from a developer point on why you don't want to develop desktop apps even if you don't want to depend on VC. This is 2009, mind you.

Desktop internal apps are a nightmare. Deployment is more difficult, it's harder to find developers willing to work on them, the only way you can access them remotely from a mobile device is using something like Citrix Server, the desktop management tools that you can use to deploy desktop apps are error prone, etc.

I said VC, but even as an architect of internal line of business enterprise apps, I would be kicked out of any managers office if I suggested developing a desktop app.

Heck these days it's hard to convince companies to write real native mobile apps instead of using one of the cross platform web based mobile app platforms.

Yes. My goal is also to make money as well as most other developers. Would you suggest I turn the focus of my career back to writing Win32 API based apps or WPF or UWP apps?

Where are all of these companies that are making money selling Windows apps? Where all of the jobs for Windows desktop developers?

In the place of myth and legend: outside of SV. I know, it's crazy.

Name some companies making any substantial money selling Windows desktop apps besides Microsoft and Adobe?

Again games don't count, they aren't for the most part developing "Windows apps" they are developing apps on top of game engines.

Autodesk, Altium, Mentor Graphics, Cadence, Keysight, National Instruments, Texas Instruments, Analog Devices, Microchip, CCS, Keil, IAR, NXP, Mikroelektronika, Mitubishi, Xilinx etc.

EDA software, CAD/CAM software, instrument control software, electronic simulation and modeling software, specialized compilers/IDEs, etc. are mostly made for Windows desktops (although more and more of them are also supporting Linux nowadays).

And if you combine the revenue of all of those companies, how do they stack up?

To put it into perspective King (maker of Candy Crush) - just one mobile app developer made $2 billion in revenue last year (https://www.statista.com/statistics/288974/king-annual-reven...)

That's about the same as Autodesk (https://www.statista.com/statistics/416235/net-revenue-of-au...).

Having a dominance in a platform (Windows) with a shrinking market is not a winning formula. Why do you think Microsoft is de-emphasizing Windows?

From no less than the CEO of Microsoft.

At our core, Microsoft is the productivity and platform company for the mobile-first and cloud-first world. We will reinvent productivity to empower every person and every organization on the planet to do more and achieve more.

You just said games don't count.

I said games don’t count as far as a reason to care about Windows desktop dominance because “Windows” games are usually written on top of a cross platform framework that is fairly easily ported to another platform. Claiming Windows is important because of the number of Unity games that run on it is about like saying Windows is important because of the number of websites you can go to on it.

If that was that simple, we would already have the year of Linux Desktop Games, we don't and never will.

Yeah, I know about Steam on Linux, just a tiny drop of Steam on Windows.

Depends on your definition of "substantial money."

There are many, many ecosystems where a Windows desktop app is the ideal platform. I spend my days writing libraries for Windows desktop software and sharing them with many other companies in the same space.

Almost by definition, most of the companies in any particular set of domains will not be household names you've heard of. There are really very few consumer software businesses, but somehow these unknown companies manage to be pretty damn profitable, even if they're not on the scope of Microsoft or Adobe.

Still no one is giving specific examples.

I spent my first 12 years as a developer doing Windows development,cut my teeth on MFC,ATL, COM/DCOM, VB6 and later on Windows Forms with .Net.

But by mid 2008 the writing was on the wall and I had spent so much time at one company, I found that my skillset wasn't marketable. No one wanted "Windows desktop developers". They wanted web developers. I've spent last 10 years being a boring old corporate Enterprise Developer nowhere near Silicon Valley.

On my list of things that I'm focusing on these days - learning at least one modern front end web framework, learning Node JS (I'm mostly a C# developer now), become more proficient with AWS from the netops/developer/devops perspective (I'm slightly above average now), Docker, etc., Windows desktop development ranks right below learning COBOL.

Looking at the market for front end technology, why would any developer who wants to stay competitive choose to focus on Windows development over Web and mobile.

Because Web never will catch up with native, is changing into yet another VM just to try to keep up with mobile, and mobile is only a money maker when doing consulting and selling training, the apps gold rush is gone.

Windows development pays the bills, owns the tablets, laptops and desktop space outside US, and if that means being paid like COBOL consultants, even better.

Do you have any numbers to back that up? The best I can find are the Stack Overflow developer survey for what developers are actually doing....


Windows software development jobs have been in decline for years...


The head hunters that fill my inbox.

I guess with 2 021 worldwide currently open positions on LinkedIn, there is something for many of us to be busy for many years.

Classic music also has been on the decline for centuries, yet composers still sell albums.

Unless "your inbox" is publicly accessible, I don't think that counts as a citation.

Working on technologies that are declining is not where most developers want to be....

It counts as much as you throwing numbers around, regardless how much you apparently hate Windows development.

Windows owns the desktop, laptops and tablets here in Europe.

Is it declining? Sure it is, consumers are no longer replacing their computers every two years, likewise with mobile phones and tablets.

Here the market speaks pre-paid, most consumers only replace their devices when they can no longer use them.

Devices around 100 euros are still being sold with Android 5.1 on them, maybe I should say Android is also declining, as it isn't profitable to spend development effort in new versions, given the adoption rate.

Android tablets are being replaced by Windows tablets across all major consumer chainstores.

All flagship Asus and Samsung tablets are Windows 10 now, that sounds pretty much desktop Windows to me.

I have seen countless technologies come and go.

Some want to be rich selling hit-Summer tunes, while others manage a comfortable life composing classic, jazz, medieval music.

It counts as much as you throwing numbers around, regardless how much you apparently hate Windows development.

The numbers I'm "throwing around" are from reputable publicly accessible sites - unless you don't think Stack Overflow is reputable. As far as the trend being away from Windows development - I posted a quote from the CEO of Microsoft in this very thread saying as much.

I don't "hate" Windows development. I would hate being stuck on unpopular, less marketable technologies. My area of expertise is still Windows back end development and ASP.Net MVC and .Net Core with a side of AWS netops, devops, and development.

Visual Studio Team Services is a joy to use for CI/CD.

I'm not disputing that Windows "owns the desktop" but the desktop market for developers is minuscule compared to web and mobile. All the money is going away from the desktop. How much money does the average consumer spend on desktop applications outside of Microsoft and Adobe? How much of B2B software as a service revenue is being spent on desktop applications compared to web applications and mobile? What are most people doing on their tablets if not MS Office and spending time on the web? Even if you're a developer, there is little money to be made creating development tools. My entire development tool chain consists of Fred Software besides the $11 a month a pay JetBrains.

Is it declining? Sure it is, consumers are no longer replacing their computers every two years, likewise with mobile phones and tablets.

We were talking about Software sells not hardware sells why are you conflating the two? As a developers selling software, The important metric is the installed based not the replacement rate. And you keep talking about the popularity of Windows desktops, but that doesn't matter as a software developer if no one is buying desktop software - except for MS Office. People are still mostly using thier desktop computers for office productivity and to go to websites.

I have seen countless technologies come and go

So have I, I got my start programming in 65C02 assembly language on an Apple //e. But I've never been one to mourn a technology that was passed its prime. That's the reason I've been moving away from the Microsoft stack toward cloud, web, and eventually mobile.

As far as where the money is for development....

Everyone including Microsoft sees the writing on the wall.



Games are one of the few remaining pieces of software that require tight platform integration. And even this is changing with WebGL.

If you take games totally out of the equation, App Store profits drop substantially. Most third-party apps are games or native wrappers for web UIs (pinterest, spotify, netflix). Of those, most are free or tied to a subscription that's device independent.

Even if that's the case, that still supports my original point that Microsoft having 90% of the desktop PC market has become increasingly meaningless in the age of web and then mobile. In the grand scheme of things, few new software initiatives are targeting Windows compared to web and mobile and only a select few companies are making any real money writing desktop software.

The example given of Autodesk kind of proves my point. They have the same annual revenue making CAD software that King digital has making Candy Crush.

Games aren't tightly tied to a single mobile platform now - most of them use a cross platform game engine. The reason they won't go to the web is payments are frictionless on mobile and you don't have to trust anyone but the platform provider.

Today anyone can setup their own shopping site and probably manage their own shipping but most small merchants still sell on Amazon.

I mean, I agree with your argument -- I'm just extending it to say that no platform dominance matters in the era of the web. Any software with mass-market appeal is going to be web-based so that it can reach the largest possible audience.


SAP isn't "windows desktop software". This whole thread started by saying that Microsoft can leverage its Windows dominance.

SAP can either be used as web based software and even thier "Windows" Software is mostly built using thier proprietary cross platform ABAP platform.

I wasn't completely serious about SAP, just meant to gesture to the the general direction of where I see commercial windows software. There are many corporate software vendors I deal with that sell windows software, for an example a suit of Office plugins that connects to SharePoint and do documents versioning.

Another area where I see windows desktop software, and I don't see it goes anywhere soon, is in industrial, medical and labs equipment control.

Sarcasm: 'Overwhelming'

> In the grand scheme of things. No one cares about desktop development.

Maybe no one in reality-distortion field of SV cares about Desktop, but offices are full of them and they're still running regular old desktop applications to do real work.

What are those regular old desktop apps? Apps built by Microsoft, Adobe, and a few other large companies. The bespoke internal business apps have been moving to the web for decades.

In SV maybe.

I have been developing mostly new WPF enterprise applications in the last 4 years and there is plenty of work across European business.

While I've not done corporate desktop development in nearly 10 years now, but have been doing enterprise apps for most of that period. Not in SV, not even in the tech hotspots of the UK, and it's all been web.

I'm not saying my experience is typical, but I suspect the bulk of enterprise apps are now browser rather than desktop based.

I've been on the opposite coast in a major metro system for the last 20 years. I've been an "Enterprise Developer" for the last 10. The number of companies that are wanting new programs that are not either web or mobile based is slim.

The focus of Google making ChromeOS desktop friendly with Android native apps, and exposing the underlying GNU/Linux userspace shows they see pressure in desktop market.

Another example is how across Europe, Windows 10 2-1 convertibles are basically wiping out Android tablets at consumer stores.

The focus of Google making ChromeOS desktop friendly with Android native apps, and exposing the underlying GNU/Linux userspace shows they see pressure in desktop market.

That has nothing to do with the statement that no one cares about Windows desktop development. All of the energy in the software space is web and mobile. Google wants to leverage the Android market - which dwarfs the install based of desktops to make Chrome OS more appealing. Apple just announced they are doing something similar - porting iOS frameworks to MacOS to make it easier for iOS developers to write Mac apps. No one would ever say that the Mac desktop market is vibrant.

Another example is how across Europe, Windows 10 2-1 convertibles are basically wiping out Android tablets at consumer stores.

Android tablets are horrible experiences. What are most people doing with thier desktop computers? Using the web and the same apps from the big companies - Microsoft and Adobe.

I think you'd be surprised how few companies that aren't SV startups actually give a damn about the cloud. "Oh, you want to charge me a subscription to use software I already own and host all my data and critical infrastructure somewhere where I don't have direct control over it? And it costs more? Where do I sign up!?"

The last company that I worked for was the stodgiest, old school non SV company that you could find.

They paid software as a service companies for everything from project management, human resource related software, mobile device management, email, they had Office 365, etc.

Most new internal software is being written using web technologies. Even if they are hosting the apps themselves. They want thier employees to be able to access their bespoke apps remotely without logging into a Citrix Server.

This is disingenuous, the big software outfits have entire teams of people to ensure their price points are reflective of existing market offerings. They'll raise prices when they can, and slightly lower them when they're looking to make a play like this. Nothing about the cloud makes this dance any different, we're still using the same steps and beat, just to an internet jukebox.

VCs only invest in “startups”, i.e. companies which can be very big very fast, and can be sold for lots of money. This does not mean that every other form of company or product is irrelevant.

If you had a great idea for a non game piece of software, what market would you target? Windows, Web, or mobile?

Where are all of the mythical profitable companies making commercial desktop software? Where are all the companies making new internal line of business desktop software?

Much manufacturing and warehouse software is on Windows, with extremely high end costs. ERPs, some are moving to the cloud but so much is still hosted locally. With HUGE annual license fees (often in the millions per year).

These systems are slow, unintuitive, and annoying. But they're stable. Companies don't want to risk their existence to change something that sort of works, so you'd be amazed at what a ERP solution with $X00,000 annual costs can get away with.

(In turn, they are often managed by / developed by the IBM/Oracle level of consulting b.s., which further drives up the costs.)

How much of the software that is hosted locally is web based software? My point is not that everyone is moving to the cloud, my point is how many new initiatives are being targeted at the "windows desktop"? Microsoft's only leverage was customers dependence on Windows APIs and fat client Windows Software.

Isn't most of Oracle's desktop software written in Java and doesn't it run on anything that runs Java?

But to answer the parent poster's question, if you were developing a new piece of manufacturing/warehouse solution, would you write it as a native windows app or would you target web or mobile? Is the claim that manufacturing/warehouse software is and will be on windows... forever?

Of course they wouldn't. They would either do it on the web or mobile. My first .Net job was writing Windows mobile apps for field technicians and factory workers on ruggedized devices. I've since written software for field service workers for Android tablets. They definitely weren't going to be carrying around Windows laptops.

>Sorry, what? Windows still has > 90% market share on desktop. If that isn't "user-based leverage" I don't know what is.

I thought that was addressed with this line:

>Windows remains an important platform in the enterprise and for gaming (although Steam, much to Microsoft’s chagrin, takes a good amount of the platform profit there), but the company has no platform presence in mobile, and is in second place in the cloud.

The 90% of users are largely enterprise (who don't go out and buy software on a whim) and Steam customers. Anecdotal, but the only reason I still have a personal Windows machine is to access my huge Steam library. I think a lot of gamers are in a similar situation. If my entire library worked on another platform (and all my preferred hardware), I would have very little reason to stay on Windows.

Keep in mind that Windows as a platform is fading away. The Web with SaaS and is taking a lot of that. Even if they keep 90% of the desktop market, revenues are migrating to cloud and apps. MS doesn't want to end up with just an OS running a browser to access other things.

Does MS not actually want it or is it your thinking?

Oh, they really don't want that because it makes the OS valueless. If the only desktop app you're running is a browser, you might as well be using ChromeOS.

But it doesn't suddenly make Windows any less valuable than ChromeOS. Which means that under the assumption of 50% of the market still using Windows, they will continue using Windows.

You may be drinking the kool-aid, but 90% of the market doesn't care. Windows costs $20 per device; ChromeOS has a non-zero cost either.

The profit margin on most consumer PCs is less than $20.


Look at the 4th chart down. That's a bit out of date, but current figures aren't going to be dramatically better, and look at the trend line.

Yeah, but Chrome OS is also developed by Google - a revenue-seeking company.

So you're either missing the point where the margins on the laptops are already including the cost of the OEM OS priced at an even lower price than a personal OEM license, or you're missing the point where the OS being free also includes your annual usage of Google services.

I’m not sure I’m following you. Bottom line is, if a consumer purchasing a laptop doesn’t care about Windows apps, two very similar laptops one running Windows and the other $20 cheaper running ChromeOS is not a good situation for Microsoft. No matter what Googoe’s revenue model is.

Well, I don't see any features of Windows that make it inherently not-free. So what's the catch here? I think people who are selling Windows believe they can milk that cow for some time, still.

During that time they can push on the software development tools/environment; and frankly they have always been better at that. Business logic is better done by business people and no matter how hard an average software developer is going to push against that - that's the truth.

Who do you think is going to acquire AirTable or alike? I could see Microsoft being incredibly interested in that opportunity.

MS definitely doesn't want it. ChromeOS has already won in terms of device administration, because simpler devices are easier to administer, less prone to malware, etc.

If native apps, e.g. Office, really get replaced by web apps, it's going to be all over for Windows as a profit center, and Windows will be hamstrung by it's legacy customers from embracing simpler ideas.

ChromeOS is no where to be seen in Europe, that doesn't sound like a victory to me.

> Seriously? GitHub could have easily gone out for more VC.

Making it as an independent company generally involves not taking on more debt on top of a massive pile of debt.

What's necessary is, you know, profits and stuff.

1) Equity isn't debt.

2) Profits? You must be kidding. Take a look at some SaaS companies that have gone public in the last 5 years. Few are profitable. It's all about growth of revenue. The profits are way down the line.

> Take a look at some SaaS companies that have gone public in the last 5 years. Few are profitable. It's all about growth of revenue.

Who's kidding? I doubt you genuinely believe we've reached a new paradigm where bankruptcy is no longer a thing.

Correct. I don’t believe most of these businesses are viable long term. But that hasn’t kept me from profiting on the stocks...

Good for you.

But you were ostensibly addressing the idea of making it as an independent company - not whether and when revenues and profits might suddenly matter again.

I believe that Github could've done their own IPO at some point. They could've "made it" as an independent company, in that respect, without profits...

And after the proceeds from the IPO have been spent?

The great think about equity is you can keep issuing more of it, doing secondary offerings, diluting existing stockholders, and raising more money until your stock price has gone to zero. That takes decades.

>Few are profitable. It's all about growth of revenue. The profits are way down the line.

We are officially well on our way to 1999 bubble territory! Nothing matters but revenue growth, even if you're selling dollars for 50 cents.

I remember 1999. We’re not there yet. Let me know when the pets.com socket puppet shows up.

.. yeah but we make it up on volume!

Equity is absolutely a form of debt, especially when given to the kinds of VCs that demand high exits.

Maybe if there were VCs who would be happy to own part of the company and be satisfied with the dividends from Github's revenue, it'd be a different story.

Maybe startups should go out and get true debt financing, then they don't have to give up that valuable stock? Oh, wait a minute, they generally can't because they have no real assets to borrow against. No, equity is not debt.

Solid analysis as always.

As a Windows user, I'm finding it increasingly harder to justify using the OS. Pretty much every single software I use to get my work done works in my browser.

The only software that runs natively is Ableton, which works perhaps even better on Mac. If I haven't switched, it's simply because of laziness: it would be too much of a hassle to copy all my samples to another device and reinstall all my plugins

The bigger challenge, if you're a long time Windows user, will be overcoming your own muscle memory. Apple's keyboard layout and shortcuts are similar but different enough to cause problems. It can be quite distracting for instance if you're in the zone and hit the shortcut to replace some text, and then suddenly every window on your desktop disappears!

Biggest problem I have with a mac is the " mark. On a PC (GB keyboard) it's shift-2, on a mac shift-2 is @, and " is shift-'

You can customize your key mappings. I found a dvorak keyboard with esperanto key bindings on the internet. If that exists, I'm sure anything else can be done (assuming you touch type).

Sure, but if you work from machine to machine to machine customisation is fairly terrible.

The author talks about leverage an OS provides to the vendor: iOS for Apple (and its App Store) and Windows for Microsoft. It is only logical that when your leverage fades, you should find an alternative means for growth.

Microsoft has been betting on the cloud for some time now. Part of that strategy is shifting enterprise from premise to Azure. In parallel investing in developers. VS Code, .NET Core and now GitHub are good examples of those investments.

What I haven't seen are non-enterprise developers, even hobbyists, using Azure as their cloud. When I talk to fellow devs, they have experience with AWS not Azure. My impression is that the Linux/CLI environment is much more familiar that Azure cloud portal with buttons à la Windows. Less black box. Curious to know if my impression is right.

The hobbyiest/ side project angle of this is the most interesting to me.

Most side projects end up in github at some point (especially if Microsoft makes private repos free - very likely). Azure’s PaaS offering is actually pretty good so having a simple deploy to azure button when creating the project could be a huge boost.

That’s my expectation, too: they don’t change anything except that there’ll be a super easy path to deploy on Azure and probably some nice CI tools, etc. with great VSCode integration. (“You have a Dockerfile. Click here to deploy on Azure free tier.”)

Simply making sure they’re considered along with AWS would be big, especially since experience on small or side projects factors into many business decisions.

Microsoft already has free private git repos, project management, hosted build servers for Windows, Linux,Macs, etc.

It's been available with VSTS for years.

I host all my side projects on VSTS.

Free source control TFS & GIT. Free builds. Free project management.

Heck, i have a complete pipeline to build my code and deploy it to the app store.

Costs me zero per month.

Please keep in mind that is "free" (has limits), but for my usage is more than enough.

I can't do the same on AWS AFAIK or even github.

AWS has a free tier but it's quite byzantine in its rules.

Azure have support for every shell you want, the problem isn't there.

My guesses are the late entry to the market, the widespread hatred to Microsoft (it's softening, but it's still there) and the half baked state of many of the services in the platform.

Pretty Apple-oriented article.

Sure, Apple has created a platform, but at which cost? You are obliged to develop for iOS, OSX only having an Apple computer. The platform has no escape rooms.

Windows has extended the .Net to run even on Linux and you can build Windows applications from other OSes as well. The flexibility is in the tooling and MS knows it - see how dramatically improved Visual Studio in latest releases and now compare it with Xcode: what's new in Xcode 10? The Dark mode, what else?

The author still forgets that actually the ultimate gaming platform is still a PC with Windows. Mobile gaming has improved, but the best gaming experience (VR and not) runs on bare metal PCs.

Apple just decided to ditch OpenGL and OpenCL in favour of Metal - this will backfire in the near future, because developers look for easy tooling which allow them to use fewer but high quality APIs and not three thousands (e.g. Vulkan, UWP).

The purchase of github doesn't make immediate financial sense to me. This article posted a reason - to stave off Windows decline. While it is plausible, I still don't see it.

The reason? Simple. Developer mindshare. Microsoft lost it with Ballmer, and is getting it back under Satya. Ultimately Microsoft's success has always stemmed from providing platforms that support developers, and in turn they build great products people will pay for. And that drives Microsoft revenues, be it the OS, or nowadays, SaaS PaaS and IaaS (and OS).

> Developer mindshare. Microsoft lost it with Ballmer

When I think of Ballmer, the first word that comes mind is "developers" (it's also the second... and the third...)

LOL Ditto. But. Being memorable for ridiculous antics doesn't quite translate into winning hearts and minds.

I think of Donkey Kong shouting "Developers Developers Developers!" while throwing chairs rather than barrels

> Developer mindshare.

Just have to point out how fucking expensive develope mindshare is then :)

But seriously, from a business perspective buy GitHub (a company which gobbled up lots of VS money without a good business model) the acquisition doesn't makes any sense.

Which MS acquisition did make sense? Skype? Nokia? LI? Yammer?


I have literally had a mobile team threaten to quit if we adopted Xamarin after they built a small app in it lol.

Xamarin can be pretty tough if you don't have a good appreciation of Xcode, iOS and UWP development and tooling, and underlying differences.

Oh, they understood it, top tier mobile devs, they just thought it was trash. Drastically preferred to do native. I went to bat for them, and we shipped a beautiful app on time.


In my view it's a longer term proposition to gain more customers for Azure/Microsoft Cloud. Google, Microsoft, and Amazon all see the future of computing as being in the cloud.

If Microsoft can integrate Azure and GitHub to make it easier for developers to deploy and run their code in the cloud, then that means more Azure customers.

This likely means that GCP and AWS integrations will take second place.

That doesn't make sense to me. GitHub integration is already pretty easy, the GitHub API is fine.

Heroku for example is already pretty well integrated with GitHub, and they didn't have to buy anything.

Maybe you know next year it will by coincidence get harder to integrate with Heroku. If they would buy it, somehow it could get harder to integrate with Azure.

They already do so.

They had a joint presentation at BUILD 2018, how to have a full CI/CD pipeline in Azure, with code taken from Github.

Now in retrospective, maybe the talks were already ongoing.

To be fair though, that sort of thing has already been very possible for anyone with the desire and engineering resources. The only thing those types of demonstrations and announcements tend to actually do is announce/release a tool that does most of the legwork for you.

Same here. I think there's another reason that's not so apparent

Cultural reasons maybe. The Ruby/Rails community Github is also part of fits into the overall strategy of cultural change towards a more design and UX centric brand, that is recognized and respected by other designers beyond being just a tool for CAD modeling.

Here is what I don't get. It is far easier for github's dev community to move to an alternate than Facebook users. Unless there is revenue to back it up, I don't understand the 7.5 Billion price tag. Heck .. Docker for 10 Billion might have been plausible. What are the barriers to another entrant?

This acquisition reminds me of the Minecraft acquisition. It did not make financial sense to me either. If Nadella keeps this up, he will eventually have issues with shareholders.

This acquisition was essential free.

This time, really, the profits don’t matter: Microsoft is paying by issuing about 73 million new shares of stock, which cost it nothing. (It’s a tiny dilution, given the company’s 7.7 billion shares outstanding; what’s more, the share price rose on the news, which means that existing shareholders are happy to be diluted.)


Using overvalued stock could make the acquisition “cheaper” than the sticker price (not really free). However, they’ve said they’ll do extraordinary buybacks to cancel the dilution in six months following the transaction. So unless the price of MSFT shares has a large correction during that period they are really going to spend over $7bn in cash.

Just to add to your response and making assumptions about how you came up with the $7 billion....

The new stock is about 1% of the total number of shares outstanding. Thier total market cap is around. $785 billion. If they did a stock buyback to cancel the dilution it would be 7.85 billion.

My numbers are rounded - the new stock is actually less than 1% and the market cap as of right now is a little less than 785 billion so around 7 billion is more accurate.

I just rounded down (to give a conservative estimate) the $7.5bn I've seen everywhere. If they are giving them new MSFT shares worth $7.5bn, buying those shares back would cost around $7.5bn assuming the price doesn't move. Of course it may be significantly more or less that depending on the evolution of the stock price.

By the way, I said they will buy back stock in the next six months but it will be in the six months following the closing of the transaction (which is expected to take place by the end of the year). I've edited my previous comment slightly.

That seems like spurious logic, especially given that Microsoft has spent many, many billions of their cash hoard on a stock buyback program (as does Apple, etc). By the broken logic of that Slate article, they're throwing money down a well foolishly because shares outstanding are "free".

Further, noting the current day price change is always the basis for countless nonsense articles. The shares haven't been diluted yet -- not until the deal closes later in the year -- and a temporary blip one way or another is close to meaningless.

Remember that AOL was able to buy Time Warner with inflated stock. If you have inflated currency (I.e. stock), why not use it to buy some real assets?

Whether a share price is inflated or not is always a point of contention -- if it's so obvious we can all buy our put options and retire on our riches. However Microsoft could literally have sold $7.5B worth of new shares and given that money to charity (which, in turn, would have been a nice tax benefit).

That money is very real, and there is nothing free about it.

Whether a share price is inflated or not is always a point of contention -- if it's so obvious we can all buy our put options and retire on our riches.

"The market can stay irrational longer than you can stay solvent".

He tripled the stock price. I doubt shareholders are sad :)

Brutus said eventually, and that's an entirely reasonable statement. Microsoft has burned many, many billions on foolish ventures to try to regain namespace, while at the same time the entirety of their revenue is still based on the coasting remains of the empire they built a decade ago. Microsoft's revenue has stagnated, despite endlessly acquiring and adding new verticals to the stack.

To put it another way, simply doing the SAP and squeezing Windows + Office would have netted a much more profitable business than everything Microsoft has been doing.

Azure is a new business.

Azure has negligible profit (if any), and it's profoundly telling that Microsoft always hogties Azure numbers with Windows Server sales. The press remarkably buys this hook line and sinker, repeats this Intelligent Cloud prattle as if it's true.

[As an aside, Microsoft's continuing tendency to group success with losers, shuffling them around to opaque the numbers, is something that sees far less skepticism than it should. We saw this with Windows Mobile cum Phone where they tried to put on the face of success for as long as possible]

Microsoft was traditionally a very high profit margin company, and their software offerings still are. Azure might sound good to make them seem like they're still with it, but it is deadly long term (massive capital expenses, fast depreciation, and very low margins).

I don't understand the last sentence:

"That, though, is exactly why Microsoft had to pay so much: buying in directly is a whole lot more expensive than using leverage, which can produce equivalent — or better! — returns for much less investment."

I think his point was that Apple has leverage over developers (you have to go through the app store to access iOS users) and therefore doesn't have to care about investing in a good developer experience. Microsoft doesn't have equivalent leverage to get developers to develop for Windows or Azure so they have to instead invest more in winning developers over by offering them better tools and services, hence the expensive GitHub acquisition.

My experience is that developers are hardcore in their opinion on Microsoft: they are either all in or have no interest in dealing with them.

I don't see Microsoft being able to move developers with this purchase.

I'd imagine there are a significant number of developers using VS Code and no other Microsoft products.

I thought Azure is the platform that Microsoft is planning to pull developers into?

Step 1: Buy github

Step 2: Integrate Azure and Visual Studio

Step 3: Profit

Interesting to see Microsoft playing nice with vendors. I’ve seen many play less than nice. (ISVs who host on Azure and badmouth Microsoft products)

Woah I had no idea that when you use Azure, you're morally obligated to like the company too.

If you’re in a partnership with them that both of you use as Marketing on your websites, you shouldn’t badmouth each other.

Applications are open for YC Summer 2019

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact