ARIN’s Public Policy Mailing List was abuzz last week with debate over the merits of modifying the needs-based justification policy for IPv4 address transfers. The debate was sparked by a proposal to increase the time period organizations can use to justify their need for IPv4 addresses to 60 months (5 years). Currently, organizations seeking to buy addresses must show that they will use them in the next two years. The two year time period was only instituted recently, when ARIN extended the period from only 1 year.
The 5-year proposal has vocal proponents and opponents. The debate is really between those who want to make market-based transfers of IPv4 address as easy as possible, and those who want to make them more difficult. Proponents of the change argue that most companies’ planning horizons are longer than 2 years, and thus they may not qualify for receiving address blocks even if they really do have a long-term need for them. The current policy, they argue, will push many organizations to seek IPv4 addresses outside of the ARIN’s Section 8.3 Specified Transfer regime, for example by buying them from legacy holders who have no contracts with ARIN. Opponents of the 5-year proposal want to prevent any departure from clearly demonstrable need, and fear that shortening the time frame for technical needs assessment will lead to speculation and hoarding of address resources.
It quickly became obvious that both camps need more data about address transactions to have an informed debate about what the period should be. Indeed, the cloud of obscurity overhanging IPv4 address transfers is almost shocking. To this end, we have suggested that the emerging address market should mimic other markets. In real estate transactions, the sales price, date, and identity of the transacting parties are readily available as public records. In some real estate markets, even more detailed purchase information is available, for instance pertaining to the financing involved. Similarly, stock market transaction data is also very transparent, especially when “insiders” are involved. This kind of information makes these markets more efficient and serves consumer protection purposes as well.
Unfortunately, very little transaction detail is currently available from ARIN – and as we have noted before, most of the other RIRs provide even less information. What is available is an aggregated view of transfers, a list of blocks transferred under 8.3, and the capability to search ARIN’s Whowas database to determine the date and identify the parties involved. But the Whowas capability sits behind onerous terms of service unlike, e.g., real estate transactions which anyone can access easily in most localities and share or analyze at will. No further transaction detail exists.
Most importantly, we do not know what type of contractual agreement with ARIN emerges from these transactions. This information has become increasingly important as large blocks of addresses are traded in bankruptcy court under conditions where ARIN’s and legacy holders’ property rights are imperfectly specified and individually negotiated. Such detail may seem arcane, but its availability could help transacting parties more easily determine the range of conditions under which a transfer could occur.
If you don’t think this is a problem, just look at some of the court documents surrounding 8.3 transfers. According to the 8.3 policy, any organization that purchases IPv4 addresses from a legacy holder is supposed to sign a standard Registration Services Agreement (RSA) contract with ARIN. But the first Nortel – Microsoft deal broke that policy. Microsoft was offered a Legacy Registration Services Agreement (LRSA). Moreover, it was not offered the standard LRSA published here, but one that was customized to its situation. No one but ARIN and Microsoft knows what is in that contract.
There is plenty of ambiguity surrounding ARIN’s relationship with purchasers of legacy blocks using 8.3. The Nortel I Order notes that Microsoft entered into a LRSA, although it clearly preserved the rights of Microsoft to do what it will with the blocks. The Borders Order similarly notes the “Purchaser’s agreement to sign a Legacy Registration Services Agreement (the “LRSA”) with ARIN.” The subsequent Teknowledge Order does not mention a LRSA, instead stating “nothing in this Order is intended to relieve or exempt the Purchaser from compliance with the policies of the American Registry of Internet Numbers (“ARIN”).” More recent Nortel II Orders (involving CSC Holdings, Bell Aliant, and Vodafone) also do not mention a LRSA, but specify that “the Purchaser(s) interest in the Legacy Number Blocks shall be subject to the terms and conditions agreed to by ARIN and the Purchaser, including applicable ARIN policies as published on its website”.
The increasingly ambiguous language in these documents suggests a disorganized market, with sellers, buyers and ARIN all trying to find agreeable transaction terms. No one knows exactly what their rights are, institutions for governing the transactions still in flux, no two transactions are comparable. At the very least, regional internet registries should all record and publish the date of the transaction, the names of the transacting parties, and clearly specify what type of contract the purchasing party is governed under. Indeed, it is hard to understand why all LRSA’s and RSAs should not be standardized. Both buyers and sellers might also be helped by knowing the price at which IP addresses are trading. As has been shown in other markets, more transparency around transfers would help reduce uncertainty and increase efficiency.