Independent axes for centralization / decentralization of transaction systems
in the never-ending struggle for sales, customers, better pricing, lower purchase costs, and better terms of trade.
Terms of reference: the transaction system of a person or business might be viewed as whatever software product(s) they use to add, edit and manage their repository(ies) of resource entries and balances. Resources include inventories and stuff, as well as external financial balances against banks and other people or businesses.
Most of the academic /whitepapers about Peer to Peer have combined various aspects of centralization into a single concept “P2P”.
Examples of some storage topologies in economic transactions:
You have various types of data:
1. Data representing the transaction amount, date, etc.
2. Indexes, pointers, and references to the transaction data
3. Control information (technical) (keys, certificates, hashes)
4. Control information (business) (control totals, balances, fiscal position)
5. Documentation/registry of technical metadata ( document schemas and
data dictionary-the meanings of data types)
6. Repository of meanings of codes (e.g. DUNS, EAN/UCC or local/bilateral)
These various data might be located on:
A. Local Machine
B. Intermediary ASP / XSP
as client-server
as publish subscribe
etc
C. Counterparty’s Machine,
D. Community business process,
E. Freenet/Cloud,
…
Overview of topologies for code execution:
Any given software module can be “located” on any of 3 or more tiers.
1. execution of business logic (business rules)
2. execution of DML (data manipulation logic)
3. population of screen widgets with information
4. delivery of User Interface
These might appear on:
A. Local CPU
B. One-tier file server architecture
C. 2…n tiers of Client-Server architecture
…
More independent dimensions in data storage and code execution topologies
Who controls technical design decisions over storage?
1. Data representing the transaction
2. Indexes, pointers, and references to the transaction data
3. Control information (technical) (keys, certificates, hashes)
4. Control information (business) (control totals, balances, fiscal position)
5. Documentation/registry of technical metadata ( document schemas
and data dictionary-the meanings of data types)
6. Repository of meanings of codes (e.g. DUNS, EAN/UCC or local/bilateral)
These might be found within:
Local Machine (Operating System vendor)
Local Machine (Application Server/ vendor)
Local Machine (Application Developer)
Local Machine (User)
Middle Tier (Operating System vendor)
Middle Tier (Application Server/ vendor)
Middle Tier (Application Developer)
Middle Tier (User)
Server/Portal (Operating System vendor) etc.
(Application Server/ vendor)
(Application Developer)
(User)
Overview of some routing topologies…
Technical Directness between self and counter party
- one hop
- thru multi-hop but dumb bridges/bit pipes
- thru Internet routing hierarchy
- thru mesh networks (community wireless networks, etc.)
-thru PSTNs
Business Directness
-thru technical intermediaries (proxies, NATs, SMTP/NNTP hosts, etc)
-thru business intermediaries (portals, auctions,
transaction notaries, etc)
1. Routes taken by data content across networks (P2P variations)
2. Routes taken by indexes, pointers, and hard metadata pointing to
the transaction
3. Routes taken by control information (keys, certs, hashes, checksums,
control totals, balances)
These can be seen in:
A. Single Hub and Spoke models,
B. Controlled Hierarchy-many levels
C. Pure P2P
D. P2P with one level of hubs
E. Managed mesh
F. Automatic mesh in the silicon
etc.
Overview of some motivation/incentive topologies…
There is always asymmetry of benefit at every decision point in every purchase and sale.
1. Capital structure (pooled/central, decentralized, or hierarch. steps)
2. Incentive topology
These can be seen in:
A. Pure anarchy models,
B. Orderly anarchy (individual sovereignty)
C. Partnerships/ joint ventures/ collaborations
D. Direct public/sector Government command hierarchy
E. Traditional Corporate command hierarchy
Incentives brain dump:
- Money: Corporate/ hierarchic (monolithic revenue capture,
central planning, central resource administration)
- Money: MLM (multilevel marketing), Network marketing, hierarchic
- Money: Pure individual / small business buy/selling
- Abstract: supply push (Sales push; web storefront)
- Abstract: demand pull (Buyer publishes requirements,
eBay search, or reverse auction/priceline)
- Volunteer motivations
- Mojo/ technically enforced motivations
There is always asymmetry of benefit in the incentives of initiator to
create electronic messages, and of recipient to respond.
Where N=2 these are not simple but have been long examined in trad.
economic theory and game theory, and reflected in artifacts like the PO.
Where N=3 or more, agent or intermediary has usually high incentive,
simplicity of incentive.
Overview of some political/force topologies…
Market, Contract, and Statutory constraints on what transactions are
allowed vs prohibited:
1. What human-level rulemaking institutions or frameworks
control the technical architecture?
2. What are the actual, operative, machine-level rules?
3. Where does actual, operative control reside — in hub,
or in the nodes? (actual sovereignty)
4. Who makes the rules?
5. Who decides upon (or constrains) the semantics used to describe a
transaction (the privilege of authoring) (incl. technical metadata
like document schemas)
These might be seen in:
A. Orderly democracy/ individual sovereignty models,
B. Pure anarchy/ gang control
C. Rule of Law
etc.
In summary, whenever a particular technical or functional requirement is served most economically by centralization, a transfer of business advantage (asymmetry of benefit) needs to be avoided by some means.
How can a transfer of business advantage be avoided, while still achieving particular technical or functional requirements that are best served by a central server?
There seems to be at least three approaches to avoid a transfer of strategic or business advantage (asymmetry of benefit):
1. Cryptographic protocols that enforce roles, protect levels of disclosure, etc. with mathematics. For example, commercial parties might publish different prices, products, or terms to various classes of customers on an open website, but provide different encryption keys to each customer or class of customer to control their access to information. The need for unix-style permissions, SSL connections, etc. is overcome. The webserver can be outsourced without loss of control of your pricing or customer list. Transaction histories can be published the same way as product list data; a single publish-and-subscribe server could stream transactions for an entire supply chain, providing very granular control over views to multiple parties such as upstream suppliers relevant to any particular transaction.
2. Statutory or contractual rules and covenants (or possibly reputation frameworks) governing behavior,
3. Market mechanisms enabling pricing to equalize the asymmetry of benefit in some classes of interactions. e.g. the individual or organization possessing particular information at particular levels of granularity might sell it for ranges of prices, thus replacing all of the previous, dumb “all-or-none” security models.
So that’s the question: how do you achieve the ultimate goal of making the network and computer do useful things, by taking advantage of its communication and computational ability without transferring advantage from one party to another? Is it inevitable that some heisenberg principle apply??
I think cryptography will have a great potential to extend the life, and the usefulness, of central servers by centralizing all the work while leaving personal choice, privacy and strategic position where it belongs — in the hands of the individual! We’ve been manipulating information for personal gains, for a long time.
The most rapid adoption of web services is enabled when you create an architecture that mines every possible economic benefit from central hosts, while at the same time, nondisruption of the pre-existing topology of strategic advantage.
