Warning (or threat?) prices must go up or .org will suffer DAYS of downtime

Public Interest Registry’s new commercial owner will have to raise domain prices significantly, or .org web sites will suffer over three days of downtime every year, one of its subcontractors has warned.
The claim came in a surprising, confusing letter (pdf) to ICANN’s top brass from Packet Clearing House, a major provider of DNS Anycast services.
PCH claims that Ethos Capital, which is in the process of buying PIR from the Internet Society for $1.135 billion, can only make a profit on the deal if it significantly ups the price of .org domains while simultaneously cutting infrastructure spending.
But its numbers don’t make a whole heck of a lot of sense to me, unless you interpret them as a threat to throw .org under a bus.
PCH is a non-profit company in the business, partly, of selling DNS Anycast services. This is the technology that allows domain names to be resolved by a server as close to the end user as possible, cutting down on internet travel time and load-balancing resolution across the world.
For 15 years, it has been providing such services to Afilias, which is the back-end registry services provider for .org and hundreds of other TLDs. Some of the money PIR makes selling .org domains therefore flows from PIR to Afilias to PCH.
While PCH is hardly a household name, even in the domain name industry (in almost 10 years, I’ve mentioned its name once), the letter, sent last week and published by ICANN last night, attempts to open the kimono a little to reveal how much it costs to reliably resolve a major gTLD.
According to PCH, “annual operational cost necessary to ensure the reliable and performant availability of .ORG” has grown from $11 million in 2004 to $30 million today.
Does that mean Afilias pays PCH $30 million a year to help resolve .org? No.
PCH says that in 2019, $1.3 million will come “indirectly from .ORG registration revenue”, with the remaining $29 million “met through tax-deductible contributions from PCH’s many donors”.
As a non-profit, PCH accepts donations from more than 30 listed sponsors, including Afilias and ICANN, as well as household names such as Amazon, Google and Netflix.
According to PCH’s letter, if .org is transferred into for-profit control, this $29 million will dry up. The letter states:

Under IRS tax law, tax-deductible donations to non-profits cannot accrue to the benefit of a for-profit. Therefore if .ORG is transferred to a for-profit entity, we cannot ask our donors to continue to subsidize its operation, 96% of .ORG’s current operational funding will disappear, and the reliability of its operation will sink from that of .COM and .NET to the least-common-denominator of commodity domains, which generally suffer several days of outage per year.

It estimates .org’s potential downtime at 3.12 days per year. It’s not saying that would happen in one big 72-hour chunk, but it still averages out at about 12 minutes per day
This amount of interruption would put PIR firmly on ICANN’s naughty step when it comes to the registry’s contractual uptime commitments — it has to provide 100% DNS service availability every month, under pain of losing its contract.
But why would those PCH contributions dry up?
Is PCH seriously saying that its donors are chucking in $29 million a year specifically to subsidize .org resolution services? Why on Earth would they do that, when .org brings in revenue of over $90 million per year and PIR only pays Afilias $18 million for registry services?
PCH provides Anycast for 243 gTLDs and 120 ccTLDs. The vast majority of these are managed by for-profit entities. There simply are not 243 non-profit gTLDs out there. Not even close.
In fact, most of the gTLDs PCH serves appear to be for-profit Afilias clients, including many dot-brands.
Goodness knows how PCH segments its income and expenditure, but it seems very likely that PCH’s donors are already financially helping to provide resolution services for commercial registries.
Could we interpret this letter as a threat to deliberately degrade .org’s performance, should the Ethos transaction go through? I’m not sure, but I think it’s a plausible read.
Regardless, we have to take PCH’s claims about the loss of sponsorship money at face value if we want to follow the rest of its calculations.

If the .ORG domain is sold for USD 1.135B, wholesale price and number of domains remain unchanged over the remaining nine years of the delegation (USD 900M gross), and operational reliability is maintained (at a cost of USD 270M), the buyer would take a net loss of USD 470M, or -6.33% CAGR. Private equity does not purposefully enter into loss-making deals. We may therefore conclude that the above scenario is not the intended outcome of the proposed sale.

That calculation seems to assume that PIR/Ethos/Afilias picks up the slack caused by the loss of the purported $29 million subsidy, rather than continuing to pay $1.3 million per year.
But PCH goes on to calculate that Ethos could make a profit on the acquisition only if it raises prices at over 10% a year AND refuses to chip in the missing $29 million.

If the .ORG domain is sold for USD 1.135B, prices are increased by 10% annually (USD 1.357B gross), and operational spending is slashed by 99%, (USD 2.7M), the buyer would make a net gain of USD 220M, or 1.99% CAGR, while increasing down-time to more than three days per year.
1.99% CAGR is not a return for which private equity would typically take this magnitude of risk. The unavoidable conclusion is that any private equity buyer who spends $1.135B to buy the .ORG domain must not only increase prices by more than 10% annually, but also cut operational costs to the minimum levels we see available at the low end of the market, with disastrous consequences for .ORG registrants and the public who depend upon them.

Again, all of these calculations appear to rely upon the notion that $29 million of voluntary donations from Amazon, Netflix, IBM, et al disappear when the acquisition is finalized.
It’s difficult to say how much PCH spends on its DNS infrastructure across the board, or how it accounts for its donations. The company does not make any financial information available on its web site.
Wikipedia reports, in an edit apparently made by PCH executive director Bill Woodcock, that the company had revenue of $251 million last year.
I assume the vast majority of that comes from and supports its primary business, which is building and maintaining internet exchange points around the world.
The only 990 tax return I could find for a “Packet Clearing House” in the San Francisco bay area shows an entity with barely $2 million of revenue in 2018.
To return to the letter, PCH concludes:

Three days per year of interrupted communications for millions of not-for-profit organizations would unacceptably damage the stability and functionality of the Internet, and more broadly of society globally.
We believe that stability and functionality should be central to any consideration by ICANN of change of control or contract modifications in relation to the .ORG TLD. As we demonstrate, the proposed transaction, or any financially-similar one, guarantees a disastrous effect on stability. Please do not approve it.

It’s a pretty shocking request, coming from an organization with a 15-year relationship with .org.
Perhaps PCH is concerned that PIR, under new management, will dump Afilias as back-end provider, leading to a loss of business for itself? Maybe, but that only appears to be a piddling $1.3 million out of a $251 million budget.
A more pressing question is arguably whether ICANN, which is currently probing ISOC and Ethos for additional information about the acquisition, finds PCH’s arguments persuasive.
ICANN has so far proved unresponsive to community concerns about pricing, but technical stability is its absolute raison d’etre. If there’s any risk at all that .org will start regularly missing its uptime targets, ICANN is duty bound to take those concerns seriously.

Tweet

Tagged: .org, acquisition, anycast, dns, ethos, ICANN, isoc, packet clearing house, pch, pir