Comcast Gigabit Upgrade Plan Sheds Light on Network Economics

Comcast’s plan to offer all its consumer customers gigabit speeds over hybrid fiber coax networks has strategic implications for cable operators, compared to most telcos and other independent Internet service providers using either fiber to neighborhood or fiber to home networks.

At the risk of seeming overly bearish about prospects for fixed network telcos, consider one key difference between the hybrid fiber and copper networks used by both telcos and cable companies, in recent decades.

Assuming a typical 500-home serving area from a single optical node, Comcast uses a fiber to neighborhood design that is structurally similar to the fiber to neighborhood design favored by AT&T, for example.

A typical fiber to the node architecture might have a single optical node serving a “few hundred” homes, making it roughly comparable to Comcast’s access network, in terms of use of optical fiber for transport and then copper for distribution to customer locations.

The difference in end user capacity is related to the differences between coaxial cable and twisted pair copper in terms of bandwidth, plus the difference between cable “broadband” and telco “baseband” modulation methods.

The differences are huge, allowing Comcast to upgrade with slight, if any changes, to the distribution plant, and upgrading to a gigabit by switching to new modems. A telco fiber to neighborhood network cannot upgrade that much by switching out customer premises gear.

So cable has a clear advantage, in terms of scaling investment to reach higher bandwidth.

Of course, the alternative is to replace the hybrid network with an all-fiber approach. Comcast itself has chosen to do so for about 85 percent of its customers who might prefer to buy a symmetrical 2 Gbps connection instead of 1 Gbps.

Verizon Communications, where it has installed FiOS, could, in principle, upgrade to a gigabit by swapping CPE and some optical node elements. Verizon has not said it will do so, but could, as market needs require.

As always, the business model is decisive. Verizon has to weigh the additional capital investment with the expected financial return. Verizon business planners might be seeing a tough case for such an upgrade, much as Verizon and others have struggled historically with the business case for fiber to the home, as well.

5G will Not Only Build on 4G, But on Cloud Computing, Virtualization and Big Data

As radical as it might seem, fifth generation mobile networks now have a design goal of eradicating the difference between access speed between fixed and mobile networks. Equally important, 5G thinking includes one millisecond end to end latency that will force a rethink of how all networks are designed and where computing activities occur.

Those requirements show how hard it will be to clearly delineate “cloud computing” from “big data” from “access networks” and “computing architecture.”

LIkewise, it will be hard to separate “fixed” from “mobile” access or “network functions virtualization” and “software defined networks” from “networks” in general.

Every one of those "silos" of thinking and practice will have to be integrated to achieve 5G objctives. At the same time, though much work lies ahead, it is possible to argue that 5G will be built on existing trends, and extrapolate logically from those trends. It is not so simple as saying "5G will be built on 4G."

It is more accurate to say that, if 5G objectives are met, 5G will be built on a variety of present trends, including cloud computing, big data, fast backhaul and fixed-mobile integration.

With the caveat that we might be overly aggressive in terms of our expectations for fifth generation mobile networks, a new International Telecommunications Union group  looking at 5G fixed infrastructure shows just how much might be changing.

ITU has established a new Focus Group to identify the fixed network requirements for 5G.

The ITU notes that its IMT 2020 program already envisions “wireless communication to match the speed and reliability achieved by fibre-optic infrastructure.”

As others note, the next generation of potential apps will require extremely low latency--with a goal of one millisecond, end to end latency as the 5G design goal.

“One-millisecond end-to-end latency is necessary for technical systems to replicate natural human interaction with our environment, a goal that experts say should be within reach of future networks,” the ITU said.

“Air interfaces and radio access networks are progressing rapidly, but there is a need to devote more attention to the networking aspects of IMT-2020,” said ITU Secretary-General Houlin Zhao.

“Today’s network architectures cannot support the envisaged capabilities of IMT-2020 systems,” said Chaesub Lee, ITU Telecommunication Standardization Bureau director.

Macrocell Leader Crown Castle Wants to Lead in Small Cells as Well

Crown Castle International Corp. (the mobile tower company)  is buying Quanta Fiber Networks (a metro fiber services provider), and the stated rationale is to bolster Crown Castle’s position in the small cell services market.

Quanta Fiber Networks owns or has rights to nearly 10,000 miles of fiber in major metropolitan markets across the United States, including Los Angeles, Philadelphia, Chicago, Atlanta, Silicon Valley, and northern New Jersey, with approximately 60 percent of Sunesys' fiber miles located in the top 10 basic trading areas ("BTAs") that arguably are top candidates for small cell networks of some size.

The acquisition of Sunesys will bolster Crown Castle's position in small cell networks by more than doubling Crown Castle's fiber footprint available for small cell deployments and expanding Crown Castle's presence in many of the top U.S. metropolitan markets.

Today, Crown Castle owns or has rights to approximately 7,000 miles of fiber supporting approximately 14,000 nodes, which contribute seven percent to each of Crown Castle's site rental revenues and site rental gross margin.

After the deal, Crown Castle will own or have rights to more than 16,000 miles of fiber suitable for supporting its small cell efforts.

That is the same sort of thinking driving Comcast’s thinking--and that of other cable TV networks--about potential uses for existing cable TV networks. Ask yourself which operators, in major markets, have dense optical fiber networks and high bandwidth nearly everywhere in a city, able to leverage those networks to support lower cost backhaul to large networks of small cells where backhaul costs are really crucial.

Looked at that way, Crown Castle will be aiming at high-density downtown areas, where cable will have the advantage outside the downtown cores.

Both strategies aim to repurpose existing assets in new ways.

Does TRAI Know Something the Rest of Us Do Not?

The Telecom Regulatory Authority of India (TRAI) has recommended that mobile virtual network operators (MVNOs) be made lawful in India. Many would note that would add yet more competition to what some would say is the most competitive mobile market in Asia.

And some observers might note that competition in India is not lacking, with some areas having seven to 11 facilities-based contestants.

Nor is it entirely clear, from the outside, what the thinking is. Perhaps there is a notion that new or niche providers could arise,as any MVNO enabling rules would allow other firms not in the mobile business to become mobile service providers.

TRAI knows full well that Indian mobile and telecom providers have not expanded so quickly in rural areas because the business model is very difficult where monthly average revenue of less than $2 a month is a common expectation.

Perhaps the thinking is to encourage cable TV operators to leverage their fixed line assets for backhaul, allowing creation of a positive business case where it does not presently exist.

Well-known brands are possible new contenders, especially in the largest urban markets, where available infrastructure and the chance to create niche products is greatest.

Observers might reasonably suggest India mobile markets are so competitive there is no room for additional retailers.

That might be quite correct for standard prepaid mobile services that sell to everyone within reach of the network.

But in other countries, MVNOs based on language, for example, have proven successful. That could well be the case in the Indian mobile market.

The recommendation goes to the Department of Telecommunications, which makes the final decision. If approved, the new rules could enable market entry by new providers in the communications business, such as cable TV operators or big brands with other assets (customers, widely used apps, distribution networks).

TRAI has recommended 10-year periods for the licenses, which could be reviewed after three to four years.

MVNOs would be required to pay government license fees and spectrum usage charges (a percentage of gross revenue) at the same rates as paid by telecom operators.

Up to this point, some marketing and branding efforts somewhat similar to an MVNO business model have been tried in India, largely without success. But TRAI understands its markets as well as any entity can. So one must assume there is a perception that under-utilized assets can be brought to bear.

Reliance Jio Hopes to Disrupt India Mobile Market

Reliance Jio Infocomm, a challenger in the Indian mobile market,  will begin offering mobile services in five Indian cities by June 2015, with a full country roll-out throughout the rest of 2015.

As you would expect, Reliance Jio Infocomm will focus first on large urban centers, including New Delhi, Mumbai, Ahmedabad, Lucknow, and Nagpur.

As an upstart, Reliance Jio will focus on the next generation of networks and services--fourth generation--in an effort to quickly climb the market share ranks. In part, that is because its spectrum is suited for 4G, and in part because 4G is an easy story to tell.

In some ways, Reliance Jio might remind some of SoftBank. In fact, Reliance Jio might not initially have wanted to both with voice, focusing instead on data services and apps. Voice is viewed as a transition strategy, giving Reliance Jio time to flesh out its “data first” strategy.

Reliance Jio also recently launched its first of several expected 4G applications, Jio Chat, a messaging application that is up against WhatsApp. Though Jio Chat does not strictly require 4G capabilities, it will be part of an application-focused service heavy on e-commerce and other apps for news, games, social and cloud storage.

The Reliance Industries is building on assets including Infotel Broadband, which had won spectrum to launch such services across India, plus spectrum purchased in the recent spectrum auctions.

The company plans to start services in 900 towns initially, according to plan. A company must roll out services across 90 percent of the metropolitan areas where it has won spectrum in an auction and 50 percent of rural areas within five years of receiving spectrum, according to the Telecom Regulatory Authority of India.

Reliance Jio has pan-India 20 megahertz (MHz) under the 2,300MHz broadband wireless access spectrum in 22 circles and 5-7MHz spectrum under the 1,800MHz band in 14 circles.

Debt burdens now will be an issue as firms divert free cash flow to paying interest and amortization on money borrowed to pay for their spectrum winnings. In fact, strategic and tactical issues during the auction might have contributed to the high prices paid for spectrum during the auction.  

Many therefore expect a wave of consolidation, at some point, that would likely reduce the number of leading suppliers from about eight to five.

Be Clear: Nobody Gets Paid for Anything When Inrternet.org is Used

Some of you likely have been wondering about Internet.org business policies--including the matter of who might pay for mobile bandwidth consumed as part of the Internet.org inititive--might exist.

Perhaps surprisingly, according to Chris Daniels, Internet.org VP, nobody is getting paid for anything as part of the Internet.org inititive.

App providers to not pay anybody to be part of the program, Internet.org does not pay any mobile service providers who participate, mobile service providers do not pay Internet.org, and the apps are available for no incremental cost to end users.

“Facebook or the developers aren't paying the operators for the data,” said Daniels. “And developers aren't paying us to be part of this program. It's free.”

“For an operator, this is a customer-acquisition tool,” said Daniels. “ It's an expense that they will take in order to bring more people onto the Internet.”

And Daniels insists the amount of incremental data usage is quite light. “The data that flows is very very light and not expensive for Reliance Communications or any of our partners,” said Daniels of its India operations.

Basically, mobile service providers and app providers are acting here on clear assumptions that everybody wins if costs do not get in the way. Over time, people who do not use the Internet will come to see the value, and eventually pay for mobile Internet access.

App providers will win as hundreds of millions of regular new users for their apps are created. People get all the benefits of the Internet one normally would expect.

And still, people complain.

Among the complaints is the limitation on bandwidth-intensive app features such as video or images.

But that just makes sense. “It needs to work for, be financially sustainable for, operators,” said Daniels. “And so, the basic services are more text-based.”

Keeping end user costs quite low is an essential requirement for bringing Internet access to hundreds of millions of new users on a sustainable basis. Internet.org helps do that, in a significant way.

Naysayers can say all they want. Internet.org helps get hundreds of millions of people exposed to the Internet, and will lead to sustainable use over time.

"When people use free basic services, more of them then decide to pay to access the broader Internet and this enables operators to keep offering these basic services for free. It is not sustainable to offer the whole Internet for free," said Mark Zuckerberg, Facebook CEO.

When "Permissionless" Innovation Potentially is Stifed by the Need to "Ask Permission"

“Permissionless innovation” is a salient talking point often mentioned in any discussion of why the “open Internet” is a good idea--until it isn’t. Whatever else one might say, banning “sponsored apps,” “zero rating” or “toll free” access definitely interferes with the notion of “permissionless” innovation.

You might say the same fundamental philosophical problem arises with any potential effort to limit, ban or prevent any compliant device or service from using unlicensed spectrum.

And that is what at least one Federal Communications Commission member warns is a problem as the FCC looks at Long Term Evolution-Unlicensed (Licensed Assisted Access) methods.

Two bureaus within the U.S. Federal Communications Commission are at the very least now seeking comment on whether Long Term Evolution-Unlicensed is actually technologically neutral where it comes to Wi-Fi access.

The whole point of having large amounts of unlicensed spectrum is promote innovation of services, applications, devices and business models.

The potential issue here is whether some Wi-Fi users (those using LTE-U) with compliant devices could be prevented from using Wi-Fi or other unlicensed spectrum using some access techniques.

That is precisely an example of “permission-based” regulation and business model policies.

As always, there are private interests corresponding to every public policy goal and practice. So long as contending Wi-Fi devices have “fair” access, there is not a problem.

What would be a problem is if some Wi-Fi devices or services--so long as they do not interfere with other Wi-Fi users in any unusual way--are prohibited from deploying an innovation that increases the value of Wi-Fi access.

Permissionless innovation is a great idea, except when the opportunity to do so is artificially restricted.

In High Speed Access, Another Turning Point has been Reached

Cox Communications says its new consumer gigabit Internet access service is now available in parts of Phoenix, Arizona; Orange County, California; Omaha, Nebraska and Las Vegas, Nevada. Cox announced it would do so a year ago.

Cox also is actively deploying “Gigablast” service to parts of Arkansas, Louisiana, Rhode Island, Oklahoma and Virginia, with service set for commercial launch later in the summer of 2015.

Perhaps more important, Cox's consumer gigabit service will be available in all of its markets by the end of 2016.

Separately, CenturyLink says it will deploy gigabit access in 2015, in addition to deployments in 10 markets, while growing the number of locations able to buy 20 Mbps and 40 Mbps by more than 45 percent over the prior year.

In conjunction with the Comcast (the largest U.S. Internet service provider by subscribers)

introduction of gigabit service to all of the 21 million U.S. homes passed by its network, the Cox announcement might represent one of those turning points we occasionally see in the communications business, namely the shift of a market from “bleeding edge” to mainstream.

It has happened before. The early leaders of the U.S. dial-up Internet access marker were not the telcos you would expect. Instead, it was new entrants, ranging from AOL in the consumer to a trio of business-focused Internet service providers (Northpoint, Rhythms Netconnections, Covad Communications).

As the market proved to be large and important, it was first telcos, and then cable TV companies, that came to dominate the segment.

As important as the Google Fiber challenge has proven to be, in terms of changing consumer expectations, it is the largest ISPs who will do the heavy lifting, as always is the case.

And that is why it matters when the largest ISPs decide they will deploy a service. No matter how many small operators decide to do something, even high take rates will not “move the needle” on adoption. Only when the largest suppliers, with the most market share, decide to move an a market really change.

That is about to happen. And the key change will not necessarily be the numbers of consumers who actually buy gigabit speed services. Most likely will not. But most are going to start buying services up to an order of magnitude faster than they had previously been using. And that is going to move the needle.

No Sign Yet of an End ot U.S. Mobile Marketing War

If you want to know how much longer the U.S. mobile marketing war can continue, the answer is simple: until the two attackers conclude they cannot take financial losses anymore.

And, so far, it does not appear that either T-Mobile US or Sprint are anywhere near that point. In its most-recent quarter, for example, Sprint added 1.2 million Sprint platform net additions compared to net losses of 383,000 in the prior year quarter.

Postpaid net additions of 211,000 compared to net losses of 231,000 in the prior year quarter. Chalk that performance up to net tablet adds.

Postpaid phone losses of 201,000 improved sequentially for the fourth consecutive quarter and improved by nearly 500,000 year-over-year.

Prepaid net additions of 546,000 led the industry for second consecutive quarter and compared to net losses of 364,000 in the prior year quarter.

So it does appear that Sprint and T-Mobile US are taking market share from the other prepaid service providers. On a net basis, it is hard to see where else the subscriber gains could be coming from.

On the other hand, wholesale net additions of 492,000 increased from 212,000 in the prior year quarter. So at least some Sprint MVNOs would seem to be adding accounts as well.

Postpaid net portings (added phone accounts less lost accounts) were  positive for the quarter for the first time in nearly three years.

Sprint platform postpaid churn of 1.84 percent improved 46 basis points (.46 percent)  sequentially from 2.30 percent churn last quarter as well, Sprint said.

T-Mobile US continued to add new accounts as well. T-Mobile US reported revenue growth of 13.1 percent in the first quarter of 2015, on the strength of robust net account additions.

T-Mobile US had 1.8 million total net account additions in the quarter, marking two straight years of adding at least a million net new customers every quarter.

Some 1.1 million net adds were branded postpaid accounts, one million of those accounts being phone accounts.

At the same time, branded postpaid phone churn was 1.30 percent.

T-Mobile revenue rose to $7.8 billion from $6.88 billion a year earlier. But heavy promotions resulted in a first-quarter loss of nine cents per share.

Oddly there is disagreement about how the marketing war is affecting AT&T and Verizon. One would think stronger customer net additions by T-Mobile US and Sprint “must” be coming at the expense of AT&T and Verizon. But that does not seem to be the case, so far.

In fact, some might argue the impact on Verizon and AT&T so far has been quite slight. Churn rates for the two carriers are stable, average revenue per account is stable and gross revenue is up, though operating income dipped, year over year.

And some would point to lower industry average revenue per account, which fell for a second straight quarter, to an average of $136/month, down from$141 in the fourth quarter.

The biggest drop happened at Sprint, where heavy promotions lead to a 14 percent dip, quarter over quarter, $132/month. But it is difficult to point to clear signs of serious financial damage at AT&T and Verizon.

One might argue it is “too early” to see the impact, but the marketing battles have been underway for more than a year. That should be enough time to discern impact, if there is any serious change.

Some might argue that most of what is happening, in terms of market share shifts, is that T-Mobile US is taking share from other prepaid services, not AT&T and Verizon.  

So how long can the war continue? Some have argued that, long term, the U.S. mobile market simply cannot support four major national suppliers. But it is hard to see, at the moment, how that consolidation would happen.

ANd without consolidation, it is hard to see the end of the current marketing wars, short of dire financial results at one or more providers.

In fact, the only scenario that would reduce immediately and clearly reduce the number of suppliers is the one development regulators will not presently support: Sprint merging with T-Mobile US, or either AT&T or Verizon buying T-Mobile US or Sprint.

In other words, none of the four leading national providers will be allowed to merge. That doesn’t mean there will not be acquisitions; there simply won’t be any mergers of the top four firms.

So, like it or not, no consolidation of the U.S. mobile market is possible by means of any mergers among the top four providers.

Instead, one of the firms would have to be weakened so much that it essentially drops from contention. That might result in a three-player market at the top, with the fourth challenger unable to compete except as a niche provider.

Right now, such a circumstance does not seem likely. Which means the marketing wars might continue for quite a longer time than some us us had expected.

High Speed Access Now is the Anchor Product for Cable TV Operators

At the end of 2014, the largest U.S. cable TV operators had about 52 million high speed access accounts in service. At the end of 2014, those same firms had some 49 million linear video customers.

In its first quarter of 2015, Comcast Internet access revenues grew 10.7 percent while business services grew 21.4 percent.

Year over year, Comcast gained 407,000 high speed Internet access customers and 77,000 voice customers and lost 8,000 video customers.

In other words, not only does the “new” Internet access business represent more customers, it also is the fastest-growing consumer service. Video subscribers actually are shrinking.

That does not mean linear video is unimportant. it remains vital, as many legacy services continue to drive the bulk of gross revenues even when in decline.

On the other hand, predictions that high speed access will become the foundation service for any fixed network operator clearly are supported by those trends within the broader industry and specifically at Comcast.

Providing Internet service providers are able to tie consumption and revenue in some relatively direct relationship, services sold to people, and based on “dumb pipe” access, will continue to grow, as a percentage of total fixed network service provider revenue.

But bundles will remain important for quite some time, because they increase perceived value, because they reduce churn and prop up average revenue per account.

About 37 percent of Comcast customers bought triple-play service. A third bought double-play services while 31 percent purchased only a single service from Comcast. In other words, about 70 percent of Comcast consumer accounts purchased a multi-service package.

About 70 percent of net product additions in the first quarter of 2015 were dual-play packages (two services), while Comcast lost about five percent net single-product accounts. Comcast had triple-play net gains of about 35 percent.

Internet.org: Sometimes Good Two Good Things are in Conflict

Right now I am trying to get some work done on a trans-Pacific flight with satellite Wi-Fi. Don't get me wrong, I'm happy to have it. But is is painfully slow. So slow it feels like less than dial-up, so I spend lots of time waiting for something to happen.

I also should add that I spend extraordinary amounts of time on mobile connections in areas with high congestion or low signal, or both. So I am used to impact of latency and rather lower bandwidth when trying to get something done.

Right now I would willingly accept a text rich but visually-limited service, with no full-motion video. And that essentially is the sort of problem Internet.org is trying to work around.

So it is with dismay that I see so much friction about Internet.org and what it is doing, on the app bundling side of the "bring Internet to everyone" effort. Yes, Internet.org is bundling apps so people with little money can afford to sample and use the Internet and some useful apps.

Nobody except potential gatekeepers likes "gatekeepers." But life is complicated. Sometimes good things are in conflict. Frankly, some of us would say a little walled garden activity or bundling, or sponsored apps and zero rating, are reasonable trade-offs for dramatically and rapidly growing the numbers of people who can use the Internet.

Some of us would say Facebook is listening. 

Apparently responding to criticism that Facebook is violating network neutraltiy principles by offering people free access (no data plan required) to bundles of useful apps, Internet.org now has opened its platform to any developers who wish to participate, with some  key stipulations.

Apparently, encrypted services will not be allowed, since all traffic has to pass through Internet.org proxy servers. Critics say the plan is “anticompetitive” because not every app, and every feature, is supported.

With all due respect, an “open” Internet is not the same thing as an “equal” Internet. When did the notion of innovation--and permissionless innovation--fall victim to the rival notion that only some types of creativity can be allowed?

To be sure, there are rival “good things” in conflict here. On one hand, we have a packaging innovation that dramatically can make useful Internet apps available to people who otherwise might not be able to use them.

On the other hand, we have the notion of fairness, that gatekeepers should not pick winners and losers. Both are reasonable principles and sources of value. But all business advantage is, at its root, “unfair.” Some products are better than others. Some innovators are more clever than others.

Sometimes you have to balance conflicting notions. That is what Internet.org is trying to do.

U.S. Mobile Carriers Expectd to Securitize Phone Installment Contracts

“Factoring” long has been a way companies can convert receivables into cash, and it appears the big four U.S. mobile carriers will use a form of factoring--actually securitizing equipment purchase contracts--more widely in the future.

All of the big carriers are likely to use equipment installment contracts as collateral for financing in 2015, Jefferies Group equity analyst Mike McCormack said. “Now there’s a cheap way to free up cash flow and use it as working capital.”

T-Mobile, based in Bellevue, may sell debt backed by customer payments on iPhones and other equipment as a way of reducing its financing costs, as it also has securitized other contracts.

T-Mobile US  weighted cost of capital is about six percent, while securitization of receivables runs about three percent, the company said.

AT&T CFO John Stephens and Verizon Communications CFO Fran Shammo have said they plan to securitize receivables. AT&T seems more specific, saying it will do so in 2015. Verizon simply says it is looking at doing so.

Sprint is likely to do so if the others move.

Jefferies analysts estimated the top four wireless carriers will help finance more than $37 billion in customer purchases throughout 2015. That may create a balance of $29 billion in cumulative receivables that could rise to $40 billion in two years.

So long as the actuarial assumptions are correct, there likely is not a problem. If assumptions about bad debt are off, the securitized loans will have the same sort of problem securitized home loans did recently.

Granted, mobile service providers and their actuarites arguably have a much better handle on account risk than mortgage lenders did. But that is not to say there is no risk.

Internet.org Goes "Back to the Future" for App Development

Lots of steps can be taken to rapidly make Internet accessible to everyone. One step you do not hear much about is “bandwidth efficiency” or "coding efficiency" or perhaps even simplicity.

But you might have heard complaints about "bloatware" or "useless features" featured as pat

As part of its creation of an Internet.org platform open to all developers, the organization argues that “to sustainably deliver free basic internet services to people, we need to build apps that use data very efficiently,” Internet.org said.

And “efficiency” will run counter to some trends common to the visual web and app world. “Websites that require high-bandwidth will not be included,” Internet.org says. “Services should not use VoIP, video, file transfer, high resolution photos, or high volume of photos.”

That focus is based on a view that networks providing very low cost or free access will have bandwidth constraints.

Operators have made significant economic investments to bring the internet to people globally, and Internet.org needs to be sustainable for operators so that they can continue to invest in the infrastructure to maintain, improve and expand their networks.

Once upon a time, all coding operated in a constrained environment, where processing, memory and bandwidth were limited. Over time, that has ceased to be a key concern for most developers.

But apps intended for use by people who cannot pay much, using networks that are bandwidth challenged, benefit from efficient apps. It has been a long time since that mattered.

The irony is that Facebook is among the popular apps that use auto-play video that dramatically boosts the amount of bandwidth consumed for use of the app.

So the issue is "appropriate technology." Where it is necessary to scale back features, perhpas that will have to be done.