This page was provoked by an e-mail thread beginning about here.
Who pays for ‘public goods’? Especially who pays for ‘non-rival’ goods which can be shared? These problems are well known to economists but lack obvious solutions. Incentives to make shared things are complex and don’t fit nicely in economic theories. Economists have argued about who pays for lighthouses since they were new technology. Indeed the only solutions I know involve either public subsidies, with their attendant political problems, or entrepreneurial solutions where someone figures how to charge for the use of shared objects. The entrepreneur must find low transaction cost tolls for access to such shared objects, for inefficient tolls mainly impact his pocket. This book tells some lighthouse history. Accounting for shared objects in a computer system may eventually need to support the entrepreneur.
Keykos was designed for a commercial environment (Tymshare) where service providers would sell stuff to other service providers in a tiered cyber economy. Classic hardware memory maps can be used in many innovative ways to meter access by a program to data. Confinement can measure the amount of information that a confined client program extracts from a memory segment. Here we mention a few more that were dreamt up. The possibilities are endless and user mode code in Keykos can implement them. Services that are had by calling objects are easier to account for but they are no easier to price.
The whole durability plan centered on the problem of a mission critical service relying on a shared object that the service could not itself afford to pay for. The result is that an object provider without a reputation can bond his object so as to convince the buyer that it will continue, at the agreed price, even after the seller is bankrupt. Again these ideas require no kernel extensions, but a few would help.
Most of these ideas have counterparts in the non-computer world and the proposed solutions came mainly from there.