Tweetstorm Digest: May 19, 2015

Lest you missed a bit of a @charlesfitz victory lap for Cloud City (Seattle) on Twitter:

1/ A geographic look at the Gartner Magic Quadrant for Cloud is revealing:

GeoMQ

2/ The Leader quadrant for cloud computing shall henceforth be known as the Seattle quadrant.

3/ Seattle has also annexed the best real estate in the visionaries quadrant.

4/ Yes, Google does their cloud infrastructure work in Seattle.

5/ The MQ has more companies from south of the Mason-Dixon line than from Silicon Valley.

6/ VMware (wedged in there like Oklahoma) is Silicon Valley’s champion, almost by default.

7/ The Cloud BS Brigade of Silicon Valley BigCos (Cisco, HP, Oracle) only appear in their own press releases.

Bonus: a fun timelapse of how this Magic Quadrant has evolved over the last five years.

Book Review: Whiskey Tango Foxtrot

The latest in digital technology and Big Data in particular are increasingly fodder for novelists. Whiskey Tango Foxtrot by David Shafer is a wry story of mostly earnest millennials who end up in “pitched battle with a fascist consortium of data miners”. The Google–esque bad guys are too nerdy and tone-deaf to achieve full-on Bond villain status, but Google is inexorably becoming the template for the next generation of thriller/movie villains (after The Interview, the only villains left seem to be evil corporations lest we offend even the most backward of movie-going audiences and/or their hereditary dictators, although they pale in comparison with classic villains in their menace, reach and ambition).

Shafer has some great turns of phrase in addition to exploring the dark side of Big Data:

They call it the cloud, but that’s wrong, isn’t it? Their cloud is heavy and metal and whirring.

But maybe that’s not how life works at all. Maybe you’re not supposed to put up so much resistance. Maybe a lot of that is pride and ego and pointless in the end. In which case she’d been misled by all that required reading and by the Die Hard movies.

Besides, there is the scrape of luck in everything, from the missed bus, to the dinged chromosome, to the hurtling asteroid.

“Why the hell would you be collecting shit like this?” Mark said, looking straight at Cole. “It’s public. It’s over our network. We call dibs on it.” Dibs? They were calling dibs? “But it’s illegal, to spy on people like this.” “Information is free. Storage is unlimited,” said Cole, totally unbothered. “Our privacy policy is reviewed regularly, and our mandate to collect is spelled out in the implied-consent decree of 2001. We’re just keeping this stuff safe, anyway. The other server giants have terrible vulnerabilities; they could be erased so easily.” Did he just smirk? “But that’s not really my department.”

“And what is the product, exactly?” asked Mark, a little desperately. “It’s a product and a service,” said Straw proudly. “It’s order. It’s the safeguarding of all of our clients’ personal information and assets. But it may be a while before our clients discover that they are our clients.”

“And yes, right now this part runs up against something called the ‘right to privacy’”—she made air quotes—“which is a notion that hasn’t really meant much in thirty years and means less every day. You may as well defend people’s right to own steamboats. Someone’s going to control access to all the data and all the knowledge. All of it. Everything that every government, every company, and every poor schmuck needs to get through the day. You want that to be the other guys? Once everyone’s on our network, the old, unwired world will be worthless.”

Tweetstorm Digest: January 26, 2015

Lest you missed an @charlesfitz Twitter excepting of a (paradigmatically paywalled) Goldman Sachs report “The Hardware Download” dated January 20, 2015:

1/ Missed a good Goldman Sachs report on cloud last week with focus on “the cloud’s impact on IBM”. Good CIO/VAR survey data.

2/ VAR survey “How has migration of your customer’s workloads to the public cloud impacted spending on infrastructure companies?”

3/ Positive responses minus negative responses: SAP +44, MSFT +40, RHT +35, VMW +6, CTXS -36, ORCL -50, IBM -74 (!)

4/ CIO survey shows $RAX a bigger player in the race for the enterprise public cloud jackpot than $GOOG

Embedded image permalink

5/ “many investors are focused on IBM’s ability to counter many of these secular pressures with its investments in cloud platforms”

6/ “the early read from our team’s surveys suggest much work is still needed in this respect”

7/ GS coyly concludes: “IBM’s infrastructure software sales could be seeing more pressure than peers with migration to public cloud.”

A Dispatch from Cloud City – 2014 Retrospective

IMG_6035

In an effort to make this an annual event, here is a plumbing-palooza stream of cloud consciousness. Last year’s broad themes remain intact though my Rackspace call didn’t pan out.

Cloud Infrastructure

  • Public cloud has won. Thanks Target. Thanks Sony. Thanks Kim Jung Un. If your public cloud gets hacked, you get to blame someone else and will have company in your misery. Public cloud will absorb vast quantities of enterprise on-premises IT spending and thus be an enormous pot of gold.
  • Docker – everyone likes Docker. Even people who don’t.
  • There are two and a half big league public cloud providers vying in what John Connors has dubbed the (WTO-free) “Battle in Seattle” (Google does a lot of cloud work in Seattle too):

Amazon remains the leader with incredible execution and is relentlessly pushing up the stack. They are on the same “enterprise journey” that Microsoft went through beginning in the late ‘90s (with some of the same people in fact), in an effort to get IT comfortable paying them vast sums of money. Amazon seems to have abandoned price leadership as they find themselves in a price war against competitors who have vastly more money than they do. Frittering away valuable cash on hardware misfires and TV shows is a growing opportunity cost. If Amazon’s stock price doesn’t recover, expect their employee retention problems to grow and discussions of spinning out AWS to get more serious. But they’re not going to yield their leadership in 2015.

What I said last year about Microsoft still works:

“Azure has become the clear challenger to AWS. The much maligned Mr. Ballmer is not getting credit for Microsoft’s embrace and execution on cloud. Unlike most of its cohorts rooted (mired?) in previous generations of technology, Microsoft is well on its way to making the cloud transition.”

Microsoft is executing like old school, taillight-chasing Microsoft with the added advantage of glass-half-full perceptions about the company for the first time in nearly two decades under the regime. The open source embrace (sans extend) is real after enactment of the strategy tax cut. If you’re still having cognitive trouble with this, the best analogy I can offer is Microsoft has become Intel and just wants to soak up all those datacenter compute cycles (a pithy analogy for what Intel has become eludes me, but it could be a fun exercise).

Google I don’t give a full big league integer to because cloud is still basically a hobby for them. In technology terms, they are in many respects the leader, but they’re just not serious about the non-technology investments they need to make to really compete for that broad enterprise transition to the cloud (they too need to embark on an “enterprise journey” as opposed to hoping those enterprises beat a path to their door). The company seems more interested in n+2 or n+3 opportunities (self-driving cars! life extension! an air force!) than mundane n+1 opportunities like cloud (which gets interesting if you believe we’re seeing weakness in Google’s search cash geyser for first time – will the further out new businesses spin up soon enough to offset slowing and/or deteriorating desktop advertising?). Presumably all those robotics investments are so they won’t have to hire humans to do enterprise sales and support. Google is the Crazy Eddie of cloud (note Eddie didn’t have much of an enterprise business and but did have a fraud problem. But far be it for me to suggest that the ad business is anything but squeaky clean). They will continue to push prices down which is a great way to push Amazon to the wall. But Google needs more than just technology and lowest price to really compete for the enterprise cloud jackpot.

  • Docker – did I mention Docker?
  • Below the big boys we have a bevy of wanna-bes, characterized by varying levels of self-delusion about their ability to really play this game. The old school announcements of “one billion dollar” multi-year investments aren’t even table stakes – Google spends that on capex in a couple weeks. IDC slyly and without elaboration predicts “75% of IaaS provider offerings will be redesigned, rebranded, or phased out in the next 12-24 months”, which brings us to this group:

IBM has been my poster child for the existential threat cloud poses to old school IT vendors. I’ve been pontificating about the peril they face and their clueless response for quite a while (here, here, here, here, here and here as a start). I took a lot of grief about this view when I first wrote about it but now their plight is widely understood and even conventional wisdom:

Bloomberg Businessweek (US)

My inner contrarian even wants to go bullish on the company just to flout the crowd except I can’t see any path that looks like clear success. Even the best outcome, where IBM keeps all its market share, still results in a dramatically smaller company (in terms of revenue, workforce and stock price) due to the deflation of cloud computing. IBM’s fundamental problem is it is their traditional customers who are being disrupted by technology wielding upstarts and they are going to have to show customers can actually use IBM technology and “business consulting” to be successful against competitors who don’t have that burden. Good luck with that. IBM’s streak as the worst performer in the Dow Jones two years running may not be over.

To their credit, IBM woke up this year and is no longer downplaying cloud or attributing their woes to simply poor execution of ye olde business model. I am amused that IBM’s leadership has expressed far more public concern about their prospects than the normally curmudgeonly IT industry analysts and pundits who evidently are telling IBM’s customers not to worry about generational transition risk.

Beyond their cloud wanna-be status, it is hard to get enthusiastic about their big initiatives of Watson and becoming an iPad reseller. After what seems like decades of hype, Watson is being devoured by hundreds of much more focused machine and deep learning startups. And it is uncanny how iPad seemed to flatline just as IBM got interested in it (and if you contend IBM’s apps are just what the iPad needs to reestablish growth, I ask only that you name an IBM app, and if you can do that, name one that you’d like to use). Delusion factor: low. The dubious marketing underscores their desperation.

HP (sorry Hewlett Packard Enterprise) trails IBM significantly in terms of existential angst and has a massive internal distraction in splitting themselves up. Helion: they only wish they had another L. While I have been assuming a “better than Autonomy” bar would lead to acquisitions like Box or Rackspace, their efforts to get their hands on VMware suggest there may be some sanity lurking somewhere. Delusion factor: medium.

Cisco is the company with the biggest gap between reality and their own cloud blather. While they are one of the few growing server vendors, their reckoning approacheth on multiple fronts. Delusion factor: highest

Rackspace – my prediction last year was they would not be an independent entity by the end of 2014. They did put themselves up for sale, but had no takers (HP let me down). They have realized they can’t play with the big boys and have retreated to their old hosting turf. OpenStack was a huge distraction for them. But their stock price supposes there is still an acquirer out there. Delusion factor: low. They touched the hot stove, and will not make that mistake again. 

Telcos – CenturyLink (also in Seattle) is executing the best here while the others are too busy chanting “cloud is our birthright” to do much. Delusion factor: medium to high.

OpenStack – another year where the number of press releases probably exceeds the largest number of nodes in production in any instance. They lost ground this year as public cloud continues to outpace private cloud and OpenStack public clouds aren’t very public. A pivot to Docker is coming, even as they perhaps settle to be a telco supplier. Delusion factor: high.

  • Docker – the most interesting aspect of Docker is it works because Linux has won as the operating system of the cloud (and having written those words, a new operating system must surely be upon us imminently). If you don’t need to virtualize multiple operating systems, you can push application isolation up above a single OS. But while Linux has won, Red Hat has lost. They just don’t play any material role in cloud infrastructure. They’re a legacy, on-premise operating system company. Maybe this year the markets will ask why they’re trading at a multiple of over 70.
  • Digital Ocean – while the old school vendors huff and puff, I’ll just note this is increasing where the cool kids run their apps. The problem with taking the “enterprise journey” is it almost always leaves you somewhere developers don’t want to be.
  • Docker – they really mishandled their first competitive blitz, which was actually pretty minimal. But good practice for when VMware finally gets around to announcing vCenter will manage both VMs and containers (and I have no inside knowledge here, it just seems like an obvious thing to do).

Cloud Platform

  • PaaS is still a zero billion dollar market, but there are signs revenue is ramping to the point where we can have a serious discussion about this threshold next year (note I define PaaS narrowly to net new, general purpose application platforms and don’t subscribe to xPaaS speciation/inclusion of decades old code) I still think this is the only layer at which most companies should be doing private cloud.
  • The Cloud Foundry Foundation – not sure if I’m more disappointed that it exists or that it wasn’t called the Foundration.
  • DevOps is a distraction and something you want other people to do on your behalf: if you’re actually doing DevOps, you’re doing it wrong. (That sentence will probably bring me more grief than any other in this post).

Big Data

  • FOMO is the biggest driver of big data in the enterprise. Lots of data is going into the lake, but not much is coming back out yet.
  • Hadoop, or more accurately HDFS, has won based on storage cost advantages and addressing the administrative and governance needs of IT. The programming model can charitably be described as unsettled, which is one of the factors hampering the realization of material value from big data. A big question for 2015 is how quickly Spark matures.
  • The most amusing announcement of the year was Google sucking it up and announcing full support for Hadoop, which they view as an obsolete and decade-old Google technology laundered through Yahoo.
  • The hype around big data will shift to the Internet of Things in 2015. IoT-washing will make cloudwashing look modest, as every data player adds at least those three letters to their home page. Some are doing more and actually building for specific IoT needs. Samsung is the biggest threat to IoT as they feel the urgency to ship half-baked spec sheets devices and crummy software that could set the whole market back significantly.
  • Get ready for data protectionism, as the EU (as a front for European manufacturers) decides they need to control their own data exhaust and not let those evil American technology companies squeeze all the value out of precision metal bending. We could see some very strange big data acquisitions by German manufacturing companies.

What else should we be watching in 2015?

Boxed In

Caged Wild Man by gillfoto, on Flickr
Creative Commons Creative Commons Attribution-Noncommercial-Share Alike 2.0 Generic License   via Flickr

The oft-repeated explanation for Box’s failure to IPO is that they somehow filed at a bad time and “missed the window”.

As usual with Box, the sound bites are better than the substance. They filed on March 24th of this year, a year that has seen the most IPOs since 2000. Since Box’s filing date there have been over 40 technology IPOs (with at least one in every month). These have included various software, SaaS and subscription companies like Hubspot, Yodlee and ZenDesk as well as HortonWorks and New Relic this week. What window exactly is closed?

Meanwhile, the stock market is not far off all-time highs, even as oil plunges, China slows, Russia invades and retrogrades, the Middle East burns, Europe self-immolates, and Ebola spreads, so it isn’t a macro problem.

The “Wall Street just doesn’t get it” argument is that because Box is a subscription business, they should get a free pass for their huge losses, because “lifetime customer value” means it will all turn out fine in the long run (this is of course counter to the Keynesian orthodoxy that in the long run we are all dead).

The people who generously have taken the time to educate us all on this topic (who coincidently often happen to be investors in Box) never seem to want to use Box’s own numbers to illustrate the virtues of a subscription business. Instead, they walk us through companies like Workday and imply that Box is somehow comparable in its subscription economics. They’re not.

There are good subscription businesses and bad subscription businesses. The spreadsheet-wielding denizens of Wall Street, despite their innumerable flaws, understand this quite well. They’ve been running the numbers on Salesforce, Workday and others for years, as well as subscription businesses in other industries.

High customer acquisition costs (traditional top-down enterprise sales model) combined with low revenue per customer (commodity file storage) doesn’t make for a good business. Add seemingly moderate churn and it can get ugly. There are some really crummy subscription businesses out there (TiVo and Vonage come to mind though it has been a while since I looked at their financials).

Add the very real cost of goods sold for petabytes of storage and the overhead of a freemium model (90% of Box customers don’t pay), and it gets even worse (I’ve wallowed in the storage economics and it can be ugly). To own Box stock, you have to believe they will retain their customers for a really long time to pay back the acquisition costs and/or significantly increase their revenue per customer. It is hard to make this case and Box notably doesn’t make much of an effort.

How will Box extract significantly more revenue per customer? They have neither moat nor unique technology (unless you count their “which one of these things isn’t like the others” participation in the Linux Foundation’s Dronecode Project). They don’t have an operations at scale cost advantage. Their “platform ecosystem” is superficial at best. They face giant competitors like Apple, Google and Microsoft with untold billions in the bank who are happily giving cloud-based storage away as a complement to their other services, as well as Dropbox which continues to ooze into the enterprise with a bottoms-up strategy which has dramatically lower customer acquisition costs. Box is still doing the same thing it always has, even as the market has evolved. They no longer have the luxury of just highlighting SharePoint’s inadequacies. Some argue Microsoft’s refusal to support Android and iOS has been the singular Box value proposition – obviously, that is a window that has closed.

The implicit financial bet/hope is Box will find a new and better business soon. But if you read the S-1, you discover Box doesn’t really know who they are or where they are going (though they do claim they’ll be really agile getting there). They’re happy to make a few jokes and compare themselves to all of Apple, Facebook, Zappos and Salesforce with a straight face, yet can’t describe their own raison d’etre (beyond giving good soundbite which I yield to no one in my respect for but tragically that competence is not a foundation for spending hundreds of millions of dollars a year):

We design our software with the passion and attention to detail that you’d expect from leading consumer companies like Apple. Similar to Facebook, we release updates to our product continuously, which allows us to act on user feedback to improve the Box experience and respond to opportunities with agility. We support our customers with the greatest care and attention, delivering Zappos-like support. And we’ve created an open ecosystem much like salesforce.com [sic], leveraging the talents and skills of tens of thousands of developers outside of our corporation to build value on Box.

What would it mean for someone to describe their company as “like Box” in a positive way?

Which brings us to their updated S-1. The good news is Box has reduced their burn, but with it their growth. Box has been a poster child for the recent “excessive burn rate” startup critique. But this is not entirely fair, as Box’s massive spending is not sumptuous employee benefits or other discretionary items, but rather is baked into the business model for customer acquisition.

Then there are the details of the Series F (as in “F#&ked”) financing Box raised this summer when it became clear an IPO was not in the cards while they were still burning mountains of cash. The private equity investors dictated stark terms and will get “theirs” well before any other investor.

Given the Series F investor preferences ratchet up even further if Box doesn’t IPO by July 7, expect an all-out push to get this turkey over the finish line. Amid that push to ring the bell, I hope Box’s fans will explain to us how the company will become a viable and even decent business in the future as opposed to just chanting “LTV FTW” because they really want to get it out of their portfolios. Customers and investors beware.

Segmenting Virtual Reality

When you say virtual reality today, most people think of Oculus VR. They singlehandedly resuscitated the ‘90s flash-in-the-pan that was virtual reality (with no small help from Moore’s Law). But the space is developing quickly and there are multiple segments emerging with very different capabilities, price points and challenges:

  • PC VR – the Oculus Rift DK2 headset famously utilizes mobile display technology but still relies on a PC to do the processing. The fact is it requires a very beefy, high-end PC to deliver the buttery smooth frame rates so important for great VR experiences. So while the headset is only $350, in practice you’re looking at out-of-pocket cost of over $2,000 unless you’ve bought a high-end gaming machine very recently. The Rift actually drove my first new desktop PC purchase in over five years (take note Intel and Microsoft). Ironically, even the latest generation game consoles may not have enough oomph to be competitive with PC-based VR, which explains Sony’s ambivalence around their Project Morpheus VR headset.
  • Smartphone VR – meanwhile the immense volumes of smartphone have driven rapid advances in CPUs and GPUs in addition to displays. The latest smartphones are very capable on the graphics front, though still materially behind PC GPUs (smartphones simply can’t consume the same kind of power). And as is typical of all things smartphone, there is a strong urge to get the PC out of the equation altogether. Google kicked things off with Cardboard. Even with Google Glass going the way of being “Segway for your face” (i.e. the uses are more mundane and commercial than consumer), they could not resist taking a shot at rival Facebook who had recently purchased Oculus for $2 billion. This cheap cardboard kit let you somewhat awkwardly affix an Android smartphone to your face. Suddenly overcoming their previous disdain for mobile VR, Oculus responded by partnering with Samsung (or perhaps found their display supply taken hostage until they agreed to partner) for what became Samsung’s Gear VR, a $199 headset to which you add a Samsung Galaxy Note 4 phablet. Unlike other Samsung spec-driven sprints to be first to market without pausing to entertain so much as a single use case, the Gear VR is actually a surprisingly good product (no doubt due to Oculus’ deep involvement, which also means this is another class of device where Samsung is relying on someone else for the software). It eliminates the wired tether of the PC-based headset but is inferior to the Rift in performance as well as a number of functional dimensions (e.g. no positional head tracking). The big problem is it requires a new (and probably unsubsidized) phone, so it still has close to a four figure price tag. But why require a new and specific model of phone at all? Why not let people use their existing, recent model smartphones (say iPhone 5 and up as well as comparable Android phones, which quickly gets you to hundreds of millions of devices)? I invested in MergeVR which does exactly this, getting the price of smartphone VR down to $99 for a headset and controller that work with your existing phone. The test of this whole category is whether you can you run a “good enough” experience on a smartphone. Because of both the Gear’s 2014 ship date as well as Samsung handing out cash to developers, the center of gravity for VR software development has shifted to the smartphone.
  • Venue VR – while smartphone-based VR solutions are driving down the price by sacrificing quality of experience, there is another trajectory moving up-market to offer the most immersive experience possible. It turns out there are all kinds of industrial applications using vestiges of ‘90s VR technology that are ripe for upgrades. These venue-based experiences tend to use the Oculus headset as well as full motion capture systems (think a dedicated room full of Kinect-like sensors) that track all your movements. With no tether, they allow complete freedom of movement, unlike the headset-only solutions which while immersive for your field of view, suffer from an “arms and legs” problem (it is like you’re sticking your head through a hole into another world while the rest of your body remains awkwardly seated). The team at AtomicVR has built an incredible venue-based system that is far and away the most immersive virtual reality I’ve experienced. The freedom of motion is amazing. You put on the headset and you’re transported into another world (beware: there are drones out to get you…). These venue-based solutions have a wide range of entertainment, commercial and industrial applications and will sell for upwards of hundreds of thousands of dollars.

I am amazed at how fast this market is moving. Oculus, the uncontested leader in VR just a few months ago, backed by billions of Facebook’s dollars, is already in a bit of a strategic quandary. They have acknowledged PC-based headsets are too expensive for a high volume consumer product, while reversing their earlier antipathy to mobile-based VR. The Gear VR partnership with Samsung halves the cost but still requires a phone that very, very few people have (or will have). After years of selling the dream of the “consumer Rift”, Oculus has gotten very vague about when if ever we will see that oft-promised device. I suspect they are seriously contemplating an OEM business model whereby they license their technology to any and every interested manufacturer on the planet as opposed to building their own devices. The division of labor and branding around the Gear VR certainly supports this theory. Oculus has incredible talent and has set an extremely high quality bar for themselves as well as their competitors (they don’t want inferior products ruining the category for everyone). The risk is “good enough” smartphone VR trumps Oculus’ perfectionism.

Thanks to Nat Brown, Mike Lenzi and Franklin Lyons for reviewing a draft of this post.

A Dispatch from Cloud City

A few end of the year observations from Cloud City (aka Seattle):

Cloud City, Bespin by TK769

Image via TK769

Cloud Infrastructure

  • AWS remains a beast. Yet a chink in their armor is emerging…
  • Azure has become the clear challenger to AWS. The much maligned Mr. Ballmer is not getting credit for Microsoft’s embrace and execution on cloud. Unlike most of its cohorts rooted (mired?) in previous generations of technology, Microsoft is well on its way to making the cloud transition.
  • Despite very strong technology and an impressive operational footprint, Google Cloud Platform is still a hobby for Google. They are as yet unwilling to make the necessary non-technology investments to really compete to win here.
  • Private infrastructure clouds just aren’t happening – instead enterprises are both getting more comfortable (surrendering?) with public cloud and continuing to invest in virtualization (VMware obituaries were definitely premature).
  • OpenStack’s identity crisis is warranted. Without a credible ecosystem of OpenStack-based public cloud providers and little enterprise private cloud adoption, the OpenStack bandwagon is left providing ingredient technology to the industry itself, which doesn’t really need what the vendors are selling.
  • Rackspace’s OpenStack bet outcome is increasingly clear: they may not exit 2014 as an independent entity. They should have invested up the stack in higher value services like Amazon, not down (and to add insult to injury, I’ll wager their VMware business still is bigger than their OpenStack business). They’ve lost over half of their market cap this year. While they still sport a premium multiple, the overall trend is towards a SoftLayer kind of valuation which could put them in play for acquisition by the kind of legacy vendors who confuse hosting with cloud (isn’t HP out telling the world their balance sheet has finally recovered from Autonomy?).

Cloud Platform

  • PaaS is still a zero billion dollar category, but could PaaS end up being the level at which enterprises implement private cloud? I see more traction for PaaS than IaaS in the enterprise.

Big Data

  • In the absence of a strong set of customer successes, I think Hadoop may be spending some time in the trough of disillusionment. The challenge is not filling the data lake, the challenge is extracting meaningful and material business results from the lake. It is a data science problem far more than an infrastructure problem. How long will it take to transition to a Hadoop 2 that is robust, deployable, performance, has ecosystem support, etc.?
  • I continue to be amused that Google is so far ahead when it comes to big data that it is a material disadvantage for them. They get dismissed as proprietary while the rest of the industry is enraptured with Google’s technology from two generations back that has been awkwardly laundered through Yahoo.

What else is happening below the clouds on the ground?