Skip to content

Commit

Permalink
Cutting loses on data for the Property and Contract chapter
Browse files Browse the repository at this point in the history
Hope to circle back with more later.
  • Loading branch information
GlenWeyl committed Mar 11, 2024
1 parent 22e3a09 commit 645f1a7
Showing 1 changed file with 8 additions and 9 deletions.
17 changes: 8 additions & 9 deletions contents/english/04-04-property-and-contract.md
Original file line number Diff line number Diff line change
Expand Up @@ -62,29 +62,28 @@ Contrast this to today's world of personal computing, where most people in devel

Much of the limited computation power then available was wasted in idle time and the feedback desired by users did not require a full machine at every desk. Instead, every user could have a basic display and input station ("client") connected via a network to a central machine ("server") whose time they shared, a set-up first pioneered a few years earlier in the Plato project at the University of Illinois Urbana-Champagne as a computer-based teaching system. This allowed ARPANET members, such as Douglas Engelbart, to simulate a future of personal computing in the era of the mainframe.

What amazing future could we simulate if we could more effectively share our computational assets? NEED A STORY HERE DRAWING ON QUANTITATIVE ESTIMATES BELOW. 1-2 PARAGRAPHS. MENTION HOW CLOUD IS A PEEK AT WHAT IS POSSIBLE.
What amazing future could we simulate if we could more effectively share our computational assets? It is hard to know without a tighter accounting of the underutilization of digital assets than we currently have. But it seems likely that we could at least buy half a decade more of effective Moore's Law simply by utilizing more effective digital assets that lay fallow. The possibilities for data sharing are richer and potentially even more transformative. Many of the most intractable problems today have answers if the power we see being unleashed by generative foundation models (GFMs) could be applied to medical diagnosis, environmental resource optimization and more that is limited by the challenges today of sharing data across organizational and jurisdictional boundaries.


The possibilities for data sharing are richer and potentially even more transformative. 1-2 PARAGRAPHS ON BOTH CONCRETE AND BLUE SKY POSSIBILITIES IF WE CAN MAKE DATA SHARING WORK BETTER. MENTION HOW LFMS ARE A PEAK AT WHAT IS POSSIBLE.

### The state of sharing

Studies of the semiconductor industry indicate that several times as many semiconductors are used in personal devices (e.g. PCs, smartphones, smartwatches, video game consoles) as go into cloud infrastructure and data centers.[^Gartner] While there is little systematic study, personal experience indicates that most of these devices are mostly little used most of the day. this suggests that a majority if not a large majority of computation and storage lies fallow at any time, not even accounting for the prevalent waste even in cloud infrastructure. Data are even more extreme; while these are even harder to quantify, the experience of any data scientist suggests that the overwhelming majority of desperately needed data sits in organizational or jurisdictional silos, unable to power collaborative intelligence or the building of GFMs.

### The state of sharing
[^Gartner]: https://www.gartner.com/en/newsroom/press-releases/2023-12-04-gartner-forecasts-worldwide-semiconductor-revenue-to-grow-17-percent-in-2024

QUANTITATIVE ANALYSIS OF WASTE AND MISSED OPPORTUNITIES IN THESE 3 ASSETS. Asset sharing may have important implications for values such as national security and the environment. Waste of resources effectively reduces the supply of semiconductors that national security policies have aimed to maximize and, like any waste, increases the demand on environmental resources per unit of output. However, it is important to bear in mind that the sources of energy employed by distributed devices and their efficiency in converting this energy to computation may in some cases be lower than those of cloud providers, making it important to pair improvements to digital asset sharing with the greening of the consumer electrical grid. Perhaps the most important implication of digital asset sharing for security may be increased interdependence between participants in these sharing networks which may bring them into tighter geopolitical alignment, especially given the requisite alignments of privacy and collaboration regulations.
Asset sharing may have important implications for values such as national security and the environment. Waste of resources effectively reduces the supply of semiconductors that national security policies have aimed to maximize and, like any waste, increases the demand on environmental resources per unit of output. However, it is important to bear in mind that the sources of energy employed by distributed devices and their efficiency in converting this energy to computation may in some cases be lower than those of cloud providers, making it important to pair improvements to digital asset sharing with the greening of the consumer electrical grid. Perhaps the most important implication of digital asset sharing for security may be increased interdependence between participants in these sharing networks which may bring them into tighter geopolitical alignment, especially given the requisite alignments of privacy and collaboration regulations.

What is most shocking about these figures is perhaps their comparison to physical assets, which one would naturally assume should be harder to share and ensure full utilization of given the difficulty of transportation and physical redeployment. COMPARISON TO DATA ON UTILIZATION OF REAL ASSET EQUIVALENCE: OCCUPANCY RATES, USE OF PHYSICAL CAPITAL, EMPLOYMENT RATES. In short, rates of waste (effective under- and unemployment) of physical assets even close to these would be considered a global crisis.
What is most shocking about these figures is perhaps their comparison to physical assets, which one would naturally assume should be harder to share and ensure full utilization of given the difficulty of transportation and physical redeployment. When unemployment rates for workers or vacancy rates for housing rise above single digits, political scandal usually ensues; such waste is omnipresent in the digital world. In short, rates of waste (effective under- and unemployment) of physical assets even close to these would be considered a global crisis.

The key reason why this silent crisis is a bit less surprising than the figures suggest is that these purely digital assets are comparatively new. Societies have had thousands if not tens of thousands of years to experiment with various social organizational systems to provide for the needs of the people's within them [^DawnOfEverything]; the origins of of our contemoporay systems of property (rental systems, capital managemnet), labor, and practices that involve the abstract representation of value [^MysteryOfCapital] (with deeds, documents issued to people, supply chain tranactions, money) can be traced to certain social-psycological qualities that arose after 1000 years of cultural practices that banned cousin marriages in Christian Europe and lead to the emergence of people who were free to form new institutions and re-constitute how property was held create new types of institutions democratic institutions that didn't exist before [^Henrich]. There have been decades to figure out how to efficiently rent cars and increasingly harness digital tools to improve the sharing of these assets (e.g. ride and house sharing platforms). Digital assets, especially those in the hands of large groups of non-technical people, date back only a few decades. A crucial task before us, then, is to determine the crucial social and technical barriers to utilizing digital assets with the same effectiveness we have come to expect of physical assets.

One way to consider what stands in the way of computational asset sharing is to consider the areas where it has been relatively successful and draw out the differences between these domains and those where it has thus far mostly failed. To do so, we will run through the three areas of focus above: storage, computation and data.

The closest framework to an open standard for asset sharing exists in storage, through the Interplanetary File System (IPFS) explicitly modeled on Lick's vision and pioneered by Juan Benet and his Protocol Labs (PL). This open protocol allows computers around the world to offer storage to each other peer-to-peer in a fragmented, encrypted and distributed manner that helps ensure redundancy, robustness and data secrecy/integrity for users at reasonable cost to those who provide storage to the system. Prominent services built on the protocol include the use by Taiwan's Ministry of Digital Affairs and other governments facing strong adversaries who may hold leverage over more centralized service providers to ensure the persistence of their data and the storage market created by PL's Filecoin system to allow commercial transactions for IPFS-based storage.

Yet even IPFS has been a limited success for "real-time" storage, where files need to be stored so as to allow their rapid access from many places around the world. It thus seems to be the relative simplicity of "deep" storage (think of the equivalent of the "public storage" spaces provided as a commodity service in real life) that has allowed IPFS to survive. Even the slightly more complicated challenge of optimizing for latency has been handled overwhelmingly by large corporate "cloud" providers such as Microsoft, Amazon, Google and Salesforce. Most of the digital services familiar to consumers in the developed world (remote storage of personal files across devices, streaming of audio and video content, share documents, etc.) depend on these cloud providers. They are also at the core of most digital businesses today.

Yet even IPFS has been a limited success for "real-time" storage, where files need to be stored so as to allow their rapid access from many places around the world. It thus seems to be the relative simplicity of "deep" storage (think of the equivalent of the "public storage" spaces provided as a commodity service in real life) that has allowed IPFS to survive. Even the slightly more complicated challenge of optimizing for latency has been handled overwhelmingly by large corporate "cloud" providers such as Microsoft, Amazon, Google and Salesforce. Most of the digital services familiar to consumers in the developed world (remote storage of personal files across devices, streaming of audio and video content, share documents, etc.) depend on these cloud providers. They are also at the core of most digital businesses today, with 60% of business data being stored in proprietary clouds and the top two proprietary cloud providers (Amazon and Microsoft) capturing almost two-thirds of the market.[^Cloud]

NEED DATA HERE
[^Cloud]: https://explodingtopics.com/blog/cloud-computing-stats, https://www.statista.com/chart/18819/worldwide-market-share-of-leading-cloud-infrastructure-service-providers/

Yet even beyond the drawbacks of this space being controlled by a small number of for-profit companies, these cloud systems have achieved in many was far less than their pioneers or visionaries like Lick imagined.

Expand Down

0 comments on commit 645f1a7

Please sign in to comment.