An Economy of Godzillas: Salesforce, Slack, and Microsoft(mattstoller.substack.com) |
An Economy of Godzillas: Salesforce, Slack, and Microsoft(mattstoller.substack.com) |
I find this statement highly suspect. What kind of world do we want to live in where regulators delineate product areas.
I see Microsoft integrating Teams into their productivity product bundle as feature; not a bug. Otherwise, we run the risk of creating separate product areas and none of them working well together — it’a going to be a minor pain in the butt to email <user@outlook.com> with the pdf tchalla@ shared on Slack; deal with formatted paste/copy & whatever else issues.
I mean, products being integrated is the whole MO of companies like Apple, Tesla and literally every other company making physical things. Sure, you can sacrifice the integration for other features like breadth of choice, specialized user needs etc, but proposing regulations to keep them distinct and separate?! That modesty sound right, imho
or, have mandated open apis that _force_ products to integrate.
That's what the web is today (mostly). Links, and embedable content (like frames). APIs and data.
The reason companies don't do this - as demonstrated fairly recently by google with their removal of xmpp protocols from google chat - is that open apis prevent lock. Open apis allows others to compete, and it is not in the interest of the existing incumbent.
IBM made a crucial mistake that apple didn't make when IBM opened the specs for their IBM compatible machines and drove down the price of PCs to what you see today - otherwise, i would predict that PCs would be just as expensive and incompatible as apple computers were.
Yep, I mostly agree. Except, defining APIs is pretty much the realm of software engineering more than regulations.
I mean, we don't want to have regulations that essentially say: "all companies selling platforms (god knows how we're going to accurately define what a platform is; let's ignore that for now) must support development against open apis such as HTML 5" and then realize that 5 years later nobody wants to write apps in HTML 5 but the shiny new thing called Flash that Adobe has developed. (My memory fails me; Flash is the hot new thing right? :P)
Anyways, it feels like society has reached a point where regulators need software skills if they're going to tackle societal impact of big tech. We need super smart folks from software _also_ become super smart folks in law and become members of congress. :D Or, maybe we need congress to sign up for #learntocode. Cross-functional skills FTW!
Are you suggesting that the web is the way it is today because benevolent governments dictated it be so? Do you have any links about this?
How? Why? It was probably bundled with Office and/or Windows.
MS got news-breaking fine for far less with Netscape before. But, nobody cares in 2020.
This is not meant to be judgemental. I'm actually of the opinion this is better in general than leaving sysadmin duties to the general public, which was largely the situation in the 1990s / early 2000s. But outsourcing this to Microsoft does of course have it's downsides, like you just experienced.
Slack did file a complain against Microsoft to the European commission.
And remember that the netscape lawsuit in EU took the good part of 10 years from first complain to sentence.
Antitrust laws are too slow to be effective.
Following Microsofts own statements and that from their sales team, Teams was indeed free to lure more users.
I don't think this particular issue is a problem, but the overall structure of tech enterprises is a problem. At least when it hits consumers instead of competitors at one point.
Amazon did that by offering superior prices, customer service, product range, delivery, etc. That's only a problem if Amazon raises prices or cuts quality after it's competitors are vanquished. What we've observed is just the opposite. Amazon continue to relentlessly push lower prices, better customer experience and faster delivery.
Competition is not a goal in and of itself. It's merely a means to the end of improving consumer welfare. Antitrust law is careful about this distinction. It's not illegal to be a monopoly because you delight you consistently delight your customers much than your competitors. In fact just the opposite, restraining a superior product from growing its marketshare would be actively harmful to consumer welfare.
The antitrust way would be to have Outlook have buttons for a lot of different meeting services out of the box. Many don't require installing any software (Google meet for example). It's all just API:s.
Some people are starting to use Teams for their meetings now because some of them like it better than WebEx, but it is exactly the same amount of effort to schedule either of them directly from Outlook.
My gut tells me this is probably quite bad overall because it is de facto total market control but only for those that are either incumbents or can somehow compete with free. It smells of AT&T, Comcast, and other “locked in” vendors, aka the most loathed companies in the nation.
At a minimum it raises the bar of entry to a market.
The software competitive scene is fairly good at the moment, and the open web remains an escape hatch, but the risk of too tight integration is that startups become impossible.
I don't know how to work around the "COM problem", though - IIRC, your product can have just as deep integrations with Outlook & Office as Teams has, if you're willing to throw man-hours at dealing with the clumsiness of COM. This understandably advantages a company with lots of money to spend and direct access to people who develop the platform. Some non-MS products still manage to ship with deep integrations, but in my experience, they tend to be fragile.
That said, I feel it would be a tremendous help for both competition and integration, if some form of "can't restrict access to a service to vendor-owned clients" clause entered the law.
It's kind of my problem with products being turned into services - service providers currently have way too much power over their customers/users.
Can't agree really. On the operating system front, it's Linux-only at this point. And the "Linux community" has completed its long way to the dark side, with RedHat paying devs to subvert the portability Unix/POSIX has always stood for, with systemd and the rush to containers. For a perspective, consider Docker was originally a way to equalize distro-specific libs and quirks; RedHat and Kubernetes essentially deprecating Docker (and adding unbelievable amounts of code to replace it as infrastructure) just means a central entity calls the shots now. When the problem of non-uniform libs is entirely created by RedHat and too many distros themselves. In reality, starved core F/OSS packages haven't changed in a decade or more. But enterprise cloud deals are just too sweet, so change for change's sake it is. In that context, it makes sense that RedHat has shutdown CentOS now.
On the web, it's even worse. There are no "browser vendors" except Google and Apple left, and Google is gatekeeping so-called web standards. Recently, HTTP and DNS is up for grabs as well. Hail monopolies.
I think it is sad that Slack is loosing independence in the the market. However, I blame investors far more than the market. When I understand it right, they yearly profit is in the 100s of millions. It is a financial decision of them.
But I find the complaints against Teams to be suspect. Primarily because Teams is not adding anything new to the MS office bundle. It’s a slightly better replacement (arguably!) for Skype.
There’s barely a feature in MS Teams that isn’t, at best, an enhancement on what MS was already delivering within Skype.
What if these (somewhat arbitrary) product categories were enforced, but instead of preventing different products from integrating, we mandated they be composable over documented APIs?
That way we could get some of the benefits of integration, but without the intensely anti competitive nature of walled garden apps?
If you don't like regulators deviding, maybe we should ask users ?
Apple has their segment walled in like North Korea.
Tesla uses religion to control their followers.
I'd love to try living in this world. What are the downsides you see?
The same world where they dictate what is intellectual property
We need something like this for software products.
I agree with you on details, but with Matt on the overall sentiment.
Most thinking on monopolies is pretty flawed, based in legalistic precedent and analogy to near anecdotes. The 1998 antitrust case against MSFT's bundling of IE to kill netscape made sense in regards to precedent (eg IBM) and economic theory. It was never the primary issue irl and the court didn't fix it despite finding against msft.
I don't think a regulator or court can just pinpoint "monopolistic behaviours" and fix it. The MSFT case proves the point. Prosecutors won, but it didn't matter much. IE still dominated. Competition waned for a while. In retrospect, the whole thing seems irrelevant or trivial to MSFT's monopoly as a whole. We are not better or worse of because of the case, probably.
A recent example is Adwords' EU case. Their dominant market share in search and search ads gave them all their minor competitors' data which was used to maintain their monopoly. The court gave google a $1.5bn fine that changed nothing. Now that precedent will be used in the US & EU to go after Amazon. They clearly use their marketplace to gain data for their retail business and skim any cream uncovered by their partners.
Whatever specific anti-competition behaviours prosecutors were able to uncover and prove are anecdotal. They "prove" that monopoly issues exist, but they aren't necessarily the actual problem. Just one specific symptom.
This is the problem. Monopolistic behaviours, individually are 90% under-water, ilegible, hard to prosecute and fixing the narrow issue you have identified and proved in court doesn't necessarily help. Some specific monopolistic behaviour isn't the point.
That said, I agree with Matt at the less specific level. Monopolies are a problem, and growing rapidly. I just think these things have to be addressed very broadly to make any difference. Classify monopolies as monopolies and apply specific rules to them. For example, "the right to be forgotten," cookie laws, data disclosure mandates, user generated copyright violations... all these highly relevant to a Google or FB. They're not as important, and much more burdensome, for the average business. These laws often exist for monopolies. Apply them specifically to them. A lot of the section 230 discussion should (imo) take this direction.
I would even be in favour of a monopolism tax. Own over 20% of the digital ad market, pay an excise tax on revenue. This would encourage competition and diversity, and also makes sense considering that such monopolies generate extra profits at the expense of the economy. Tax all monopoly mergers, or ban them... not specific mergers. IE, FB have a monopolistic position in social media. No mergers for you. Let those acquired companies compete with FB instead of widening its moat.
If a court/legislator is going into the nitty gritty of how the monopoly works, what monopolistic behaviours exist and such... the actions still needs to be broad. If amazon or google are using a monopoly over a (literal) marketplace to secure a monopoly as a supplier to that marketplace.... break that up. Make adwords a separate company to search. Separate amazon's marketplace business from the retail business. Don't let them create a delivery business. etc.
Trying to act narrowly is pointless. Any ruling on the nuances of how amazon uses data from their marketplace business to help their retail business is pointless. Instead, recognize that there's a structural problem. Amazon run the market and also a retail business. They're using that monopolistically. That's a bad structure, and any specific bad thing amazon does within that structure is not relevant. The structure itself is relevant. That's the whole premise of antitrust in the first place.
Predatory pricing is just dumping. You sell something below cost until your competition goes out of business, then you either raise prices or hold prices constant but reduce quality (eg Amazon selling counterfeit books).
Bundling is requiring people to purchase A+B in combination. An example would be a gourmet ice-cream maker that is actually famous for chocolate but only sells 50% chocolate, 50% vanilla.
You can see this move as a combination of bundling and predatory pricing. Microsoft did this with Defender too. Their strategy is to starve nascent competitors for revenue in a niche area so they can never mount a direct attack on the core business.
I can't think of one example of a company killing a competitor via free products and then raising prices.
And let's not pretend Amazon's counterfeit book problem was part of the business plan.
> That modesty sound right, imho
Meant to say: "That doesn't sound right, imho"
We need to find a way to discourage mega-mergers. Either through antitrust law or progressive taxation on mega mergers starting at few $1Bln.
1. The economy is a whole losses efficiency due to lack of competition.
2. Less competition for talent.
3. More fragile economy (e.g. too big to fail).
https://www.economist.com/special-report/2018/11/15/across-t...
https://hbr.org/2018/03/is-lack-of-competition-strangling-th...
Justifying the need for it, go though the paper work, estimate costs etc. VS just adding a new line item in an existing bill.
And part of the reason why I hate the trend of turning everything into services is that regular (middle-class and below) people don't get to outsource the increasing amount of relationships they have to enter into.
You have one person to talk to, they manage RocketChat, NextCloud (with an office suite, agenda, files), Jitsi, Discourse, CodiMD, all of this with a single sign on for your coworkers.
All for 10€/month/user.
It's not as integrated as say a full microsoft suite, but probably one order cheaper.
If you add these I'll have bingo:
* "We already have X for this" (X is only vaguely related and/or does it far, far inferior)
* "it's too expensive" (said by $200/hour consultant vetoing $50/month SaaS)
* "it doesn't conform to the architecture" (used for anything that isn't Java and Microsoft Office)It made a compelling case for breaking up large corporations in the interest of reducing inequality (via more competition and reduced consumer prices). However, it's the only book I've read on the subject so am unsure if it's a widely held and reliably informed view.
For commenters that are wondering what world we'd be living in with functional anti-trust laws, the book suggests we look to the US between the late 1930's and the early 1980s.
[1] https://www.google.com/search?q=site%3Amattstoller.substack....
In the netscape-msft case, prices didn't even exist. They don't exist or aren't central in a whole lot of modern businesses. How do you analyze FB in terms of the CWS model? It's like trying to examine the health of a rose bush by taking its temperature.
It's a pretty good example of intellectual failure in economics, compounded by legislative/industry agendas. Defining reality using a model rather than the other way around. The model is good at measuring market power of something like a centralized buyer of agricultural commodities. The more unlike an agricultural cartel the industry, the lower the "evidence" for monopoly.
https://youtube.com/playlist?list=PLATIVW2S3zKMbVlnAHAPR0sHZ...
Bork's tedious tome on antitrust was birthed directly from Aaron Director's Antitrust Project at the University of Chicago. (Bork credits Director fatuously in his introductin.)
1. Competition for the sake of competition is not an end in itself. Competition is a means to an end, not a promise or guarantee. See also: the T-Mobile and Sprint merger.
2. New businesses are started as new opportunities arise. Wherever a market inefficiency is perceived, there are people who seek to exploit this inefficiency for their own benefit. Even the businesses that fail employ some people for some time; it’s not important that the same businesses that exist today continue to exist in perpetuity, but that the business environment continues to encourage people to take new risks and start new ventures.
3. Stop bailing out failing businesses. Would it hurt the S&P 500 if a FAANG company filed for bankruptcy? Yes, but that’s irrelevant. Socializing losses only encourages more businesses to take on more risk than would normally be acceptable to them and their shareholders if they think there’s a good chance they’ll be bailed out by taxpayers. Sometimes businesses, even big businesses fail, and that’s part of the environment, not something to be prevented at all costs. The stock market is a reflection of market activity, it isn’t the market.
1. Not really true in general, because the buyee often was not in direct competition with the buyer. In specifics, sure sometimes the buyee is bought and killed, and sure sometimes "build it (and compete)" was actually in consideration at buyer -- though as an aside if they did build it instead of buy it we'd still get complaints from other people on finding a way to discourage product expansion and vertical integration. One could also say here that by eliminating middlemen (if the buyer intends to integrate that is) efficiency is increased, but this is not necessarily general either.
2. Marginally less employer competition, and perhaps not even that, because under 0 we've got a group of beneficiaries who may now be able to form or fund new companies, in direct competition or not. Avoid static pie thinking. Related here is an opposing concern that preventing mergers can result in a lower cap on wages, though this is not general either. I do concede that if we expand the pool of employees beyond software engineers, some sort of layoff protection may be desirable, because technically and instantaneously yes the now-irrelevant laid off accountants for example at the buyee have a tighter market to compete in as buyer doesn't want them and their old job has vanished. (Though many layoff packages already implicitly do something like this, the net effect is to give the individual more time to find a role at a different firm or else retool because the market has reallocated the demand for their role.)
3. Being too big to fail isn't really related with the scope of a company's product catalog. Fragility is a big topic.
Rather than spouting my own ideological biases and principles I'd rather ask some numbers questions. Specifically what thresholds would you consider reasonable before introducing your discouragements/blockers. #2 seems most tractable to start with -- say the US has 5 million software engineers, Facebook has 50,000 of them, and wants to acquire another company with 15,000 software engineers. Is the shift above 1% of employees in the field to 1.3% objectionable? Or do we not worry until a single firm employs say 30% of the field? Is the relative increase of 0.3% too big a merger regardless of the starting point? Or is there even such a too-big relative size and we should just look at the final total? Should we count the total employees at each company, or rather subdivide even further taking into account sales, accountants, and perhaps even types of software engineers?
1) you arent charged if youre a writer, de facto.
2) yes, they provide stuff that costs money if you want to use them (custom subdomains cost a one time fixed cost of $50)
3) you, as a content creator, can choose if you want to enable monetization, at which point the readers must pay to read your content, but still - you arent explicitly charged to post stuff (which is what happens in the Medium model, implicitly).
one thing Substack should do is enable webhooks so that if you are running a webapp and you want to give your paid subscribers access to an 'exclusive blog', you can tell substack that user X with email addr Y is a paid subscriber to MyService ABC, and then instead of requiring payments to Substack, it would just let them through. Not sure how the user/unit economics would work here, but I enabled a Substack blog for a new SaaS I'm building, and I want to make it pay-only, but I'm already using Stripe to handle subscriptions/payments and I don't want my users to have to pay twice in order to see the blog (while already paying for the APIs and services provided by the site.) and also use the service web app as well as the service web API.
@cjbest, @sethbannon, @pg if youre listening I think this would be a good thing to prioritize: webhook to allow subscribers for X's SaaS/startup/whatever to be implicitly include in the set of "paying Substack users" instead of requiring some app's user to pay a second time (for some non-egregious fee, maybe a small percentage of the user subsription fee for X's pricing model. Maybe even Stripe & Substack could chat about a bidirectional integration to make this seamless, and even bake in some of the costs).
if you like these ideas you should totally reach me at the addresses listed on my keybase on my profile, I am super awesome and its 4:24am EST (my location) so clearly I'm motivated.
Cheers
> Substack makes it simple for a writer to start an email newsletter that makes money from subscriptions.
Like medium, it's a way for writers to get paid. There seems to be popular tech and business writing there. HN's own patio11 is a fan.
For example, the regulator can mandate that if you have an app that you sell (SaaS or not), you must also provide a way to export the data out in a machine readable, documented format.
Regulators can mandate that your app must have an api to insert compatible data into your app (like embedding). Regulators can mandate that your app must have an api to make it possible to embed into another app, or some other requirement.
It's up to society to make these regulations, just like how society mandates building safety code, electrical safety code etc. Companies won't do this willingly of course, but they can if their competition is also forced to (level playing field).
Things are a little different now that private industry is funding most of this stuff
And they are probably happy about that.
You can also see it with in retail and agriculture. Producers get exploited in the name of the customer. Retails wants as few suppliers as possible and left unchecked, it just reaches unsustainable proportions that certainly are unhealthy in a lot of ways.
Or you end up with "open standard with proprietary extension" (bane of everything from HTML to opengl)
https://bugs.launchpad.net/bugs/1
"Microsoft has a majority market share in the new desktop PC marketplace. This is a bug which Ubuntu and other projects are meant to fix."
I think you could also avoid using the word API entirely. I could see the end result being similiar to how default app selection works in Android.
That said, I still haven't sat down and thought through the unintended side effects.
Err, the concept of containers aren't being deprecated. At first, Kubernetes only supported Docker. This was considered a negative monopoly, so they defined the CRI (Container Runtime Interface) and started shifting towards that. Now they're just taking the sensible step of deprecating the old specific support for Docker. Docker is still free to implement CRI (and IIRC there have been motions towards this). Containerd (Docker's underlying runtime) already supports it today.
Your Docker images will also keep working. Even if Docker doesn't end up implementing CRI, both containerd and CRI-O use the same container format, and are compatible with your existing images.
A GPL'd system available entirely for free is hardly a "monopoly", though. And this only applies to servers, Linux desktops are rare and phones are split between the free-ish Android (with critical non-free Google Play Services) and the entirely locked down iOS.
Yes, that would create some accounting overhead. But it's a small price to pay to keep Godzilla from eating the villagers' livestock.
A central service where you can easily manage many subscriptions, with integrations into other websites. There are a lot things I would pay for if I didn't have to go through the process of creating and managing yet another account. If it was just a one-click add/remove.
News websites come to mind. I would also use a "buy access to this article for 50 cents" feature. It just has to be frictionless.
The major players might not want to give up the money from people buying subscriptions and not using them much or forgetting to cancel them, but that seems like a lunch ready to be eaten.
There’s downsides to both.
We’re on a trend of unbundling right now, but the pendulum is quickly shifting the other way.
I believe Apple folks recently said it's actually good to have Android around, so it prevents them to be under anti-monopoly laws.
*Once you know where to look.
It's now a "Microsoft 365" thing.