New gTLD winners will be decided by lottery after all

ICANN will use a lottery to decide the order in which to process new gTLD applications, after a surprising U-turn.
ICANN this morning published a proposal that would prioritize applications based on a $100-a-ticket prize draw that would run in early December.
The results of the draw would be used to sequence applications for Initial Evaluation and, if successful, contract negotiations, pre-delegation testing and eventual delegation.
ICANN says the draw would give it an exemption to California’s anti-lottery laws, which was the primary reason it has so far resisted chance-based solutions to the batching/sequencing problem.
It’s applied for a special “fundraising drawings” license based on its non-profit status, which it expects to be granted before the end of November.
The license appears to have certain restrictions that confuse matters for applicants — they won’t be able to buy their tickets over the internet.
They’ll have to pay, in-person, for a paper ticket. But ICANN says that it can supply proxies for applicants at no cost, eliminating the need to fly a representative to California.
The whole process will be manual, so there’s little risk of an embarrassing Digital Archery-style snafu.
Applications for internationalized domain names would be given priority.
The draw would be run at some point between December 4 and 15.
Under the proposal, the results of Initial Evaluation would start to be released from March next year, starting with IDNs, at a rate of about 150 per week.
ICANN has also decided to extend the period for official objections to March 13, 2013, two months more than the current plan, due to requests for more time from potential objectors.
But the extension is unlikely to appease these objectors, which will still have to file objections before they know whether applications have passed Initial Evaluation, wasting money.
New gTLD applicants that pass Initial Evaluation, are not in contention and have no objections will have the option to immediately sign the standard registry contract.
Applicants wishing to negotiate their contracts will be processed according to their draw number.
However, no contracts will be signed before the ICANN meeting in Beijing next April. This is because the Governmental Advisory Committee does not expect to issue its formal Advice on applications before then.
ICANN expects to sign contracts and do pre-delegation testing at a rate of about 20 per week, which is roughly within the maximum 1,000-per-year delegation rate it has committed to.
The effect of this is that the first new gTLDs are expected to go live in the DNS root in the second quarter of 2013, rather than the third quarter.
I believe most of the proposals will be welcomed by most applicants. A lottery was always the most favored solution.
There will be some criticisms, however.
There does not appear to be a method envisaged for swapping slots, for example, so portfolio applicants probably won’t get to choose which of their gTLDs is delegated first.
The whole proposal is open for public comment here.

New gTLD hopefuls set aggressive targets for ICANN

ICANN should start delegating new gTLDs in the first quarter of next year as previously planned and the Governmental Advisory Committee should work faster.
That’s according to many new gTLD applicants dropping their ideas into ICANN’s apparently semi-official comment box on application “metering” over the last week or so.
ICANN wanted to know how it should queue up applications for eventual delegation, in the wake of the death of batching and digital archery.
According to information released over the past couple of weeks, it currently plans to release the results of Initial Evaluation on all 1,924 still-active applications around June or July next year, leading to the first new gTLDs going live in perhaps August.
But that’s not good enough for many applicants. Having successfully killed off batching, their goal now is to compress the single remaining batch into as short a span as possible.
The New TLD Applicant Group, a new observer group recognized by ICANN’s Registry Stakeholder Group, submitted lengthy comments.
NTAG wants Initial Evaluation on all applications done by January 2013, and for ICANN to publish the results as they trickle in rather than in one batch at the end.
The suggested deadline is based on ICANN’s recent statement that its evaluators’ processing powers could eventually ramp up to 300 applications per month. NTAG said in its comments:

Notwithstanding ICANN’s statements to the contrary, there is not a consensus within the group that initial evaluation results should be held back until all evaluations are complete; in fact, many applicants believe that initial evaluation results should be released as they become available.

That view is not universally supported. Brand-centric consultancy Fairwinds and a couple of its clients submitted comments expressing support for the publication of all Initial Evaluation results at once.
January 2013 is an extremely aggressive deadline.
Under the batching-based schedule laid out in the Applicant Guidebook, 1,924 applications would take more like 20 months, not seven, to pass through Initial Evaluation.
NTAG could not find consensus on methods for sequencing applications among its members. Separate submissions from big portfolio applicants including Donuts, Uniregistry, TLDH and Google and smaller, single-bid applicants gave some ideas, however.
Donuts, for example, hasn’t given up on a game-based solution to the sequencing problem – including, really, Rock Paper Scissors – though it seems to favor a system based on timestamping.
The company is among a few to suggest that applications could be prioritized using the least-significant digits of the timestamp they received when they were submitted to ICANN.
An application filed at 15:01:01 would therefore beat an application submitted at 14:02:02, for example.
This idea has been out there for a while, though little discussed. I have to wonder if any applicants timed their submissions accordingly, just in case.
Comments submitted by TLDH, Google and others offer a selection of methods for sequencing bids which includes timestamping as well alphabetical sorting based on the hash value of the applications.
This proposal also supports a “bucketing” approach that would give more or less equal weight to five different types of application – brand, geographic, portfolio, etc.
Uniregistry, uniquely I think, reckons it’s time to get back to random selection, which ICANN abandoned due to California lottery laws. The company said in its comments:

Random selection of applications for review should not present legal issues now, after the application window has closed. While the window was still open, random selection for batches would have given applicants an incentive to file multiple redundant applications, withdrawing all but the application that placed earliest in the random queue and creating a kind of lottery for early slots. Now that no one can file an additional application, that lottery problem is gone.

Given that the comment was drafted by a California lawyer, I can’t help but wonder whether Uniregistry might be onto something.
Many applicants are also asking the GAC to pull its socks up and work on its objections faster.
The GAC currently thinks it can file its official GAC Advice on New gTLDs in about April next year, which doesn’t fit nicely with the January 2013 evaluation deadline some are now demanding.
ICANN should urge the GAC to hold a special inter-sessional meeting to square away its objections some time between Toronto in October and Beijing in April, some commenters say.
ICANN received dozens of responses to its call for comments, and this post only touches on a few themes. A more comprehensive review will be posted on DI PRO tomorrow.

Infodump: what we learned about new gTLDs today

ICANN held a webinar today in which it detailed a whole lot of the current thinking about the evaluation phase of the new gTLD program, including some new deadlines and target dates.
Senior vice president and acting program head Kurt Pritz fought through a cold to give new gTLD applicants more information and clarification than they’d received since Prague in June.
These are some of the things we learned:

• Three applications have been withdrawn already. We don’t know which ones.
• There have been 49 requests to change applications. Again, we don’t know which ones yet. ICANN is in the process of finalizing a threshold check to allow or deny these changes, details of which it expects to publish soon.
• “Clarifying Questions” are the new buzzword. CQs — yes, they have an acronym — are additional questions the evaluators need to ask applicants before they can score parts of their application. The vast majority of applications are going to get at least one CQ. The two-week deadline to respond to these questions, as described in the Applicant Guidebook, will likely be ignored in many cases.
• About 90% of applications will get a CQ about their financial status. This mainly concerns their Continuing Operations Instrument, the super-complex and expensive back-up cash commitments each applicant had to secure. But applicants who got letters of credit don’t need to panic if their banks have recently had their ratings downgraded.
• Another 40% can expect to get questions about their technical plans. Some applicants may have relied too heavily on their back-end providers to describe their security plans, it seems.
• About half of all geographic gTLD applications have not yet supplied letters of support from the relevant government. This was already anticipated and is accounted for by Guidebook processes however, Pritz said.
• Don’t expect an answer to the metering question any time soon. Batching may be dead, but ICANN does not expect to figure out its replacement — a way to throttle new gTLDs’ go-live dates — until October. There’s an open comment period on this and plenty more jaw-wagging to come.
• Objections will come before Initial Evaluation results. This sucks if you’re a likely objector. The deadline for filing objections is January 12, 2013, but evaluation results are not expected until June 2013 at the earliest. This means the much cheaper option of waiting to see if an application is rejected before paying for an objection is no longer a viable strategy. But it’s good for applicants, which will get a little more visibility into their likelihood of success and their costs.
• Contention sets will probably be revealed in November. The String Similarity Panel, which decides which gTLDs are too similar to each other to co-exist, is not expected to give its results to ICANN until late October, four and a half months after the June 13 Reveal Day — so applicants won’t know the full size of their contention sets until probably a couple of weeks after that.
• The new gTLD public comment period will probably be extended. After several requests, ICANN is very probably going to give everyone more time to comment on the 1,930 1,927 applications, beyond the August 12 scheduled closing date. An announcement is likely on Friday.

New gTLD timetable expected this week

ICANN plans to publish a new timetable for its new gTLD program later this week, according to its latest update.
Its board of directors’ New gTLD Program Committee said in a report (pdf) published this morning:

The roadmap will show how the separate schedules for evaluation applications, possible dates for GAC [Governmental Advisory Committee] input, comment & objection periods, and other program elements fit together. The plan will demonstrate interdependencies, indicate risk areas, describe schedule uncertainty, and indicate how applicants might be affected by changes to the plan.
The roadmap will be released by the week of August 6, 2012.

New gTLD applicants have been waiting for this report since the Prague meeting in late June, when it became clear that the original timetable, based on application batching and “digital archery”, was dead.
Potential objectors will also be sharply impacted by the timetable; decisions could hit their wallets.
If the window for filing private sector objections closes before the GAC deadline to object, for example, the cheaper wait-for-the-GAC strategy for objecting becomes a non-starter.
Today’s report from ICANN also discloses a little more about how the 1,930 new gTLD applications are being processed: they’re being grouped by applicant and/or by back-end registry provider, in an attempt to create efficiencies.
According to ICANN, this will enable the evaluators to ramp up to a maximum capacity of 300 applications per month, but that it will take a few months to fully ramp up to that speed.
The Initial Evaluation phase of the process began about a month ago, in line with its July 12 target date, ICANN said.

Adding some time for ICANN to organize and publish results, this means that initial evaluation results will be published in 11-12 months after the July 12 start date, i.e., May or June 2013.

With the timetable set to be published this week, the ongoing public comment process about application metering will presumably not have an impact on what is published.
With that in mind, any timetable released this week is unlikely to answer every outstanding question about the timing of go-live dates for successful new gTLD applicants.

New gTLD winners could be named June 2013

ICANN has sketched out a tentative timetable for the evaluation of its new generic top-level domain applications that would see the first successful gTLDs appear over a year from now.
But the plan has little meat on its bones, and ICANN has admitted that it still doesn’t know exactly how the evaluation process is going to pan out.
In a new call for comments, ICANN confirmed that all 1,930 applications are going to be evaluated at the same time, and that the evaluators have already started work.
The winners and losers from Initial Evaluation, ICANN said, could be announced June or July 2013.
This would mean that the first new gTLDs would start going live on the internet “in late third quarter of 2013, six months later than originally expected”, ICANN said.
But which successful applications would start hitting the root first is still wide open to debate.
The idea that the applications would be processed in batches of 500 or thereabouts, is now pretty much dead. That’s been obvious since digital archery was killed off, but it’s now confirmed.
ICANN said it has a “tentative project plan” that “foresees the processing of applications in a single batch, and simultaneous release of results” about a year from now.
But with “batching” dead, we now have a “metering” problem.
Hypothetically, as many as 1,409 unique gTLD applications could emerge successfully from evaluation at the same time, in June or July next year.
That’s the theoretical ceiling; in reality the number will be substantially reduced by withdrawals, objections and contention.
But before any of them can go live the applicants need to negotiate and/or sign registry agreements with ICANN and undergo formal pre-delegation technical testing. That creates two bottlenecks at ICANN in its legal and IANA departments.
ICANN now wants to know how to “meter” successfully evaluated applications, to smooth out the roll-out so that no more than 1,000 new gTLDs are delegated in any given year.
An idea that emerged in Prague was to order applications according to how “clean” they were, as measured many clarifying questions the evaluators had to ask the applicants. But that idea has now been dismissed as “unworkable”, ICANN said.
ICANN’s board of directors had promised to make about three weeks after the Prague meeting – a deadline that passed over a week ago – but it’s now turning to the community for ideas.
Before August 19, it wants to know:

1. Should the metering or smoothing consider releasing evaluation results, and transitioning applications into the contract execution and pre-delegation testing phases, at different times?
a. How can applications be allocated to particular release times in a fair and equitable way?
b. Would this approach provide sufficient smoothing of the delegation rate?
c. Provide reasoning for selecting this approach.
2. Should the metering or smoothing be accomplished by downstream metering of application processing (i.e., in the contract execution, pre-delegation testing or delegation phases)?
a. How can applications be allocated to a particular timing in contract execution, pre-delegation testing, or delegation in a fair and equitable way?
b. Provide reasoning for selecting this approach.
3. Include a statement describing the level of importance that the order of evaluation and delegation has for your application.

My hunch based on conversations in Prague is that the majority answer to question 1 will be “No” and that the majority answer to question 2 will be “Yes”, but that’s just a hunch at this point.

Digital archery looked “silly” but had “minor risks”, ICANN board was told

While ICANN staff acknowledged that digital archery was perceived as “silly”, it told the board of directors that it was “straightforward” and “unambiguous and easy to execute”.
That’s according to the latest delayed release of meeting minutes and briefing documents detailing board-level discussions between the Costa Rica and Prague meetings.
There was significant debate at the board level about digital archery prior to its approval in March, with directors generally favoring an auction model instead, these documents reveal.
Digital archery as a method of batching new gTLD applications was approved by ICANN at the end of March. It was then suspended two weeks ago and finally killed off last Wednesday.
Back in March, the board of directors’ new gTLD program committee was presented with a strong case in favor of archery by ICANN staff.
According to March 28 briefing document (pdf):

Implementation of the auction model at this late date presents significant risk of: program delay, legal action and significant reputational impact as described below. Board working group members tend to agree with this viewpoint but there is a split of opinion. The digital archery model presents minor risks; primarily a minor reputational risk from the perceived awkwardness of the model.
Analysis indicates that the legal risk raised by a random selection program will be satisfactorily addressed. This is true even though the results appear to have an element of randomness.
…
Implementation of the digital archery model is essentially completed. It presents no schedule risk. Its operation is straightforward.

Hindsight is a wonderful thing, and it reveals in this case that digital archery was far from straightforward in its operation, and did in fact present schedule risk.
The new gTLD program is currently in semi-limbo while ICANN tries to figure out a way to sequence the processing of applications in a fair and timely way.
Other documents published following Prague include the lengthy minutes of a May 29 committee meeting at which directors argued with staff about how to geographically weight batches.
Staff pushed for a proportional system – where if 10% of applications came from a specific region, 10% of the first batch would be drawn from that region – the minutes reveal.
But several directors argued and won the case for the “round robin” scenario, which would have given advantage to applicants from under-represented regions instead.
Newly published minutes from May 6 also reveal that ICANN considered offering 1% interest on refunds to applicants that withdrew their applications before Reveal Day.

Digital archery is dead, but uncertainties remain

ICANN has killed off its unpopular “digital archery” scheme, which it had planned to use to rank and batch new top-level domain applications for evaluation.
But the organization has not yet replaced it with anything, leaving gTLD applicants without their much-sought-after certainty for at least the next three weeks.
In a resolution yesterday, ICANN’s New gTLD Program Committee approved the following resolution:

Resolved (2012.06.27.NG06), the New gTLD Program Committee directs the President and CEO to terminate the Digital Archery process as approved in Resolutions 2011.12.08.04-2011.12.08.07.

Given the discussions between the ICANN board and the rest of the community here at ICANN 44 in Prague this week, it would have been more surprising if archery had survived.
Not everyone is happy to see it go, of course.
Richard Schreier, CEO of erstwhile digital archery service provider Pool.com, took to the mic at the ICANN public forum this afternoon to ask that ICANN sticks to its decisions in future.
He further noted that the decision to scrap archery had been made without the input of applicants who are not in attendance at the meeting.
Now that archery has gone, the ICANN board has left a vacuum – nobody knows how applications will be prioritized for processing and evaluation.
Committee chair Cherine Chalaby said that ICANN will now open a comment period for all applicants, in order to help build a “roadmap” to “detail the next steps and timelines”.
This roadmap is due, it seems before the new gTLD committee’s next meeting, which is due to take place approximately three weeks from now.
This does not necessarily mean the program has been delayed, however. ICANN senior vice president Kurt Pritz said a few times this week that evaluators will start looking at apps July 12.

New gTLD application batching dead?

It’s looking increasingly possible that not only is digital archery dead, but that ICANN may also kill off the idea of batching new gTLD applications entirely this week.
Given the number of groveling apologies from staff and board about the failure of digital archery over the last few days, there doesn’t seem to be any way it can be plausibly reinstated following its shut-down last week.
But from the first three days of meetings and hallway conversations here at ICANN 44 in Prague, it’s pretty clear that doing away with batching is under serious consideration at board level.
It’s also my understanding that ICANN staff, which initially appeared reluctant to abandon plans to divide the 1,930 applications into three or four batches, is now also thinking along the same lines.
Batching is unpopular among most — but by no means all — applicants, because they don’t want to risk losing a time-to-market advantage by being allocated to a later batch.
When director Chris Disspain told an audience of applicants yesterday, “What I think the board thinks you want now is certainty,” the reaction suggested he had hit the nail on the head.
The problem ICANN has with a single batchless evaluation process is that it faces — hypothetically at least — up to 1,409 unique gTLDs exiting Initial Evaluation at the same time.
This could cause problems because it’s promised the DNS root server operators and the Governmental Advisory Committee that it will delegate no more than 1,000 new gTLDs per year.
These commitments are, at least for now, non-negotiable, chairman Steve Crocker has indicated this week.
So ICANN has to figure out a way to “rate limit” application processing so that no more than 1,000 gTLDs go live in the same 12 month period.
Many opponents of batching have stated that the process already contains several throttling mechanisms, or “gateways” as ARI Registry Services CEO Adrian Kinderis calls them.
For starters, not every application will be successful. Some will be withdrawn soon because they were tactical filings, others will not pass Initial Evaluation and will be withdrawn.
Some will fail Initial Evaluation and enter Extended Evaluation. Others will face formal objections or will find themselves in contention resolution.
In these cases, applicants can expect an extra six months of processing time, which will act as a natural throttle.
For those applications that get through to contract negotiations, ICANN’s legal department will operate on a strict first-in-first-out basis with the paper contracts, Disspain said yesterday.
But there’s a concern that these gateways might not be enough to smooth out the evaluation and approval process.
Various solutions have been put forward by the ICANN community so far this week.
These have ranged from the predictable “IDN applicants should go first” from IDN applicants and “brands should go first” from brand applicants, which both seem unlikely to be adopted, to some more inventive ideas.
Top Level Domain Holdings founder Fred Krueger and others have suggested that one way to prioritize applications would be to ask the large portfolio applicants — TLDH, Google, Donuts, et al — to decide which of their gTLDs they want to hit the root first.
“I value .london significantly more than I value .beer,” Krueger said yesterday. “I’m sure Google values .google more than .lol.”
Another idea, put forward by Uniregistry’s outside counsel Bret Fausett yesterday, was to rank applications according to how cleanly they exit Initial Evaluation.
Applications that made it through Initial Evaluation without the evaluators needing to ask any clarifying questions would be considered the “first batch”. Those that needed a single question answered would be the “second batch”, and so on.
This system would have the advantages of enabling a single batch while rate-limiting applications based on their inherent quality.
On the face of it, it’s quite an attractive idea, and it’s my sense that Fausett’s approach was well-received by ICANN. We might be hearing more about it as ICANN 44 progresses.

Digital archery suspended, surely doomed

ICANN has turned off its unpopular “digital archery” system after new gTLD applicants and independent testing reported “unexpected results”.
As delegates continue to hit the tarmac here in Prague for ICANN 44, at which batching may well be hottest topic in town, digital archery is now surely doomed.
ICANN said in a statement this morning:

The primary reason is that applicants have reported that the timestamp system returns unexpected results depending on circumstances. Independent analysis also confirmed the variances, some as a result of network latency, others as a result of how the timestamp system responds under differing circumstances.

While that’s pretty vague, it could partly refer to the kind of geographic randomness reported by ARI Registry Services, following testing, earlier this week.
It could also refer to the kind of erratic results reported by Top Level Domain Holdings two weeks ago, which were initially dismissed as a minor display-layer error.
TLDH has also claimed that the number of opportunistic third-party digital archery services calibrating their systems against the live site had caused latency spikes.
Several applicants also said earlier this week that the TLD Application System had been inaccessible for long periods, apparently due to a Citrix overloading problem.
Only 20% of applications had so far registered their archery timestamp, according to ICANN, despite the fact that the system was due to close down on June 28.
Make no mistake, this is another technical humiliation for ICANN, one which casts the resignation of new gTLD program director Michael Salazar on Thursday in a new light.
For applicants, ICANN said evaluations were still proceeding according to plan, but that the batching problem is now open for face-to-face community discussion:

The evaluation process will continue to be executed as designed. Independent firms are already performing test evaluations to promote consistent application of evaluation criteria. The time it takes to delegate TLDs will depend on the number and timing of batches
…
The information gathered from community input to date and here in Prague will be weighed by the New gTLD Committee of the Board. The Committee will work to ensure that community sentiment is fully understood and to avoid disruption to the evaluation schedule.

Expect ICANN staff to take a community beating over these latest developments as ICANN 44 kicks off here in Prague.
There’s light support for batching, and even less for digital archery. It’s looking increasingly likely that neither will survive the meeting.

ARI: digital archery is a lottery and we can prove it

ARI Registry Services has tested ICANN’s digital archery system and concluded that it’s little better than a “lottery”.
The company today released the results of a network latency test that it conducted earlier this month, which it says proves that applicants in North America have a “significant advantage” over others in securing a place in ICANN’s first new gTLD evaluation batch.
ARI basically tried to figure out how important the geographic location of the applicant is on digital archery.
It concluded that the further away you were, there was not only more network latency, as you would expect, but also that the latency became less predictable, making archery less about skill and more about luck.
According to the company (with my emphasis):

The conclusion is simple; the closer an applicant is to the ICANN Data Centre in Virginia, the greater likelihood of repeatable results, allowing a significantly higher chance of calibrating the network latency and thus setting a low Digital Archery time. It is therefore a significant advantage being located as close as possible to ICANN’s Digital Archery target or employing an organisation who is.
It is ARI’s contention that the frequency and size of network changes seen in networks outside North America mean the greatest influence on an applicant’s Digital Archery shot is luck. The further one is from North America, the greater the influence luck has on an applicant’s Digital Archery shot. Those applicants without the resources to access systems or representative organisations within North America are to all intents and purposes, playing a lottery, hoping that latency remains consistent between their calibration tests and their actual shot. The applicant’s ability to influence this game of chance reduces the further they are from North American networks.

While it might read for the most part like a technical white paper, make no mistake: this is a strongly political document.
By putting this information out there and linking it directly to the legally scary word “lottery”, ARI knows that it is putting ICANN in a very uncomfortable position.
The reason ICANN settled upon the digital archery system in the first place — rather than the preferred option of random selection — was because gambling is illegal in California and the organization’s lawyers were worried about nuisance lawsuits.
ARI has, essentially, just given fodder to the kinds of legal vultures that will be thinking about such lawsuits anyway.
The company is one of the strongest opponents of digital archery. In a recent interview with DI, CEO Adrian Kinderis called for batching to be scrapped in favor of a single evaluation period of 10 to 12 months.
You can read ARI’s 19-page report here.