Plenary #ifipwg82 Karin Knorr-Cetina, Keynote IFIP WG 8.2 Working Conference, Dublin 2016
Introduction by Séamas Kelly, The Centre for Innovation, Technology & Organisation, University College Dublin
This digest was created in real-time during the meeting, based on the speaker’s presentation(s) and comments from the audience. The content should not be viewed as an official transcript of the meeting, but only as an interpretation by a single individual. Lapses, grammatical errors, and typing mistakes may not have been corrected. Questions about content should be directed to the originator. The digest has been made available for purposes of scholarship, by David Ing.
Welcoming Karin Knorr-Cetina, U. Chicago, Otto Borchert Distinguished Service Professor, Departments of Sociology and Anthropology
Science and Technology Studies
Post-social relations, to broaden social inquiry, relations with objects as well as other human beings: interobjectivity
[Karin Knorr-Cetina]
Title has changed from “What If the Screens Went Black? The Coming of Software Agents”, published in the proceedings at http://dx.doi.org/10.1007/978-3-319-49733-4_1
First time to Dublin
Home country of Austria
Will talk about work on finance
3 technological transitions in finance: good this is a working conference, will discuss, they’re not finished
Developing a technosphere, looking a performance level, how is technology used
1. A form of temporalization: A new ontology, fluid and liquid
2. From technology to media
3. Scaled up social form with meta-actors
Interest in exchange markets
Concentrations in market: UK, U.S. Japan
Over the counter market: huge, foreign exchange between banks
Not regulated much, done in exchanges
Technology is trading screens
1. Streamers 1886-1980s
2. Scopes 1980s
3. Algorithms 2000s
Original stock ticker: cross between printer and telegraph
Connected exchanges
Not related to foreign exchange, but stock markets
Name of security, price change time, transaction time
People then had the market on tape
Before calling, wanted to know who was trading what at what price
Transcript from Knorr & Preda 2007, from the days in which ticker tape was still used
Saw price differentials
First transition to ticker tape had consequences over 10 to 15 years
Beginning regime of attention, so could know what was going on
Expanding information boundaries, could see not only what was happening in New York, then Boston, then London — a telegraph
Origin of a new community of practice related to ticker tape, today called technical analysts — looking at numbers from tape, and then ordering them differently, e.g. price differentials over time, volumes
New social stratification around ticket possession: in-group tried to prevent out-group from getting ticker tape
Ticker tape created a sequential time
Taking manifested in trading in market, temporalization of market
Changed with computer screens, 100 years later
Face to screen
On screen, more than price
Market world on screen: if it’s not on the screen, it doesn’t exist, e.g. 9/11 was on screen quickly
Scopic media: screen-based technologies
First Reuters Monitor, 1973
1981 action capability, could trade through the screen, didn’t need phone, went live to 145 institutions in 9 countries
At 2012: 400,000 Thomson-Reuters terminals; 315,000 Bloomberg terminals world
Had put big sheets on floors, with prices written manually
Scoping mechanism:
Instant visible making of the market that is reflexive
Watch the market on screen, but also trade, which changes what you watch
Augments: offers analysis, information, news about what’s going on, referential whole in a Heideggerian sense, to the audience
Audience responds to projected reality on screen, rather the pre-reflected (e.g. phone calls)
Why “media”?
Always an audience
Can analyze as dramaturgical
Analytical staging of the market
Transition to scopes: from piping to beaming
Piping: social relation based (networks), maintain relationships over the telephone
Beaming: tele-coordination
Despite underwater pipes (networking), performance isn’t a network anymore, as everyone sees the same screen
Tele-coordination: When you scan scope it, you don’t have to network it!
Market always has an audience
Neuro-market synchronization
Neuro-physiological level
Asked traders how they trade, not getting an answer: seat of pants, know what to do (which doesn’t explain anything to a social scientists)
Social science says something about implicit knowledge, but it’s not science
Neuro-physiology not yet a science: have pre-frontal (explicit, accessible to conscioius); implicit (is skill based, content not verbalizable, inaccessible to conscious awareness)
Two performance systems recruit and develop different cognitive circuits
Can develop pre-frontal explicit
But in market where have to respond in seconds to market changes, rely on implicit so no delays in sequence, but more traders are integrated into the scopic system
Implicit cognitive processes, can’t do by thinking, have to act
Results in a regime of attention, traders have to watch the market
System between trader and screen is false: implicit processing to the neural level
Master skiers say “You can’t win a thing with thinking”
Based on implicit processing
Explicit system is capacity limited, processing 4 to 6 variables at the same time
Making a serve in tennis takes more than 4 to 6 things
Also, it’s sequential
Get a neuro-market synchronization
Global
Everyone watching the same screen, at an implicit level
Traders give indications of that: don’t know how to trade
Ethnography, staying at distance doesn’t work
More like a tiger watching a prey, ready to jump
Feels like a physical connectedness
When the market turns against them, they feel as if they’re sexually violated, e.g. I got shafted, I got hammered
Not a psychic level, it’s emotional
Given neuro connection, maximum implicitness has negative aspects:
Task subtractions: doesn’t see the market
Implications are subtracted
Content subtracted
Social-moral subtraction
Inability of Wall Street to understand
Second transition: unified global market, 1980s-2005
Interbank trading, complete
Streaming market in analytic time
Analytic and differential time: time flow, have to get into this
Ontology of the real is slowing
Continuation of temporalization in speed: schedules, regime of attention physiological
Moving from screens to servers watching algorithms
Subterranean action of algorithms
Now observe spiking, e.g. NYSE isn’t now dominant — from 2004 80%
Rise of algorithms that trade
A major disruption
Temporalization of social world, could be at its limits in terms of speed, as at speed of light
Vocabulary changes: low latency, speed trading
Also thinking about infrastructure at speed of light
Fibre optical cable can go to 2/3 of speed of light, but microwave can get to 99%
Microwave connections (New York to L.A.) compared to fibre optic (to Japan) will gain 4 to 5 seconds
Described by Michael Lewis in Flash Boys, between Chicago and New York
Change in practice
Trading rooms use to be communications rooms (although they were through screen)
Constant talking, now not so much
Trading rooms are mostly silent
Competing on speed, rather than trading strategy
South Korea is creating barriers to increase speed further, goes back to trading strategy over speed
Humans disappear
Proprietary trading farms, as opposed to doing on public trading floor with big banks
Firms have become traders
Question on integration of non-human actions into the social world
Luckmann 20 years ago: why does the social world always have to be near to human being? It’s not necessary
Are we integrating algorithms into the social world?
Professional changes:
Desks become firms
Drastically reduced number of skilled traders, rest done by algorithms
Influx of quants (math, stats, physics)
New educational requirements on remaining human traders (who used to be apprentices in banks for 2 years before becoming assisting traders, now they need a university degree with sophisticated math)
Algorithms:
Instructions to accomplish a task
Also are programs to act: e.g. learning algorithm might have models implemented in the algorithm
Algorithm can then be a trader: assistant to human traders (in simple transactions)
Strategic counterpart into human beings, could become dominant market actor
There are firms where only algorithms change
May have to reconceptualize markets
Infrapersons:
From the technical literature on what algorithms do
Algorithm as unemotional as seen as one of best features
Algorithm may be scientific; global
Human actors have use discipline to rein in emotions
Algorithms are dumb, don’t need discipline
Human being self-regulation with moral judgement; software agents regulate via risk assessment
Algorithms are fast, unlike human beings
Develops algorithmic culture of time
Leads to flash crashes
First flash crash of May 10, 2010 (unresolved today), market crashed
Dow down 1000 points before recovering, $1 million value erase
All happened within 13 minutes, then recovering in 26 minutes to normal orderly
Stop of the market was due to a breakdown of technology: correct market prices weren’t being displayed, even though trading could happen
Have had many flash crashes:
Nobody really loses
Rebounds as quick as failing (as the market speaks)
Liquidity refreshed
October 2016, GBP mysterious crash, no explanation
2010 crash could be explained by aggregate data, done by SEC
Nothing economically intersting happens, just have to stick it out
If don’t panic, get excess turbulence, and then recovery
Is this an economic ritual?
Goffman says rituals are symbolic representations of social order
Symbolic representations of social order
Flash crash doesn’t cast a shadow
This is beyond interpretation
Economic ritualization
Market could create synthetic persons, at speed of light
Still holds up
Crashes don’t bring the market down, crashes increase solidarity with the market
Scaling the market?
Crashes could bring in a phenomenological others, encountered as an actor
Not like an automatic car (which is more a tool)
Economic ritualization have financial markets emerge from change
Conceptualization: institutions versus markets, market isn’t a hierarchy, it’s flat social form, no CEO or governance structure
Trading use to have an interactional moral level
In addition to infrapersons, have metapersons (an in between category)
CEO level is not involved in trading
Trading floor not managed from above
Traders aren’t watching trades, they’re managing and changing algorithms (at least every 3 months)
Traders aren’t called traders any more, meta-person not doing the activity
Involved in advanced cognitive: math skills, combined with engineering skills
If have meta-persons, market is no longer a flat transactional form
It has levels
Traders become an epistemic class
Beyond knowledge society, based on theoretical knowledge (Daniel Bell)
Epistemic class of sophisticated math with agency engineering and money making
Used to be acephalous social form without a master, but now masters have matured (and breed)
Knowledge society is pushing itself through, not just technical analysts, but watching and changing algorithms
Summary:
Talked about 3 disruptive stages
Looked a performative view of technology: temporalization, mediatization, rescaling of social form
Rescaling of social form into two levels: meso level on market; rise of hybrid epistemic class
[Comments by Hugh Wilmott]
Screen went black an hour ago, because the talk changed
Conference theme: shift from what technology means, to socio-technological appendices do
Examining in context of financial markets
Financial markets have been under-researched, except for people in financial economics
Few in organizational studies have researched
Some change, due to financial crisis
Important, we rely on those markets
Applications of technologies are at the cutting edge
Position with socio-technical systems to post-WWII period, could re-engage
Think of issues of power and class
Coal mining studies of early 1950s
Emerging hybrid epistemic class
Pre-reflexive interaction
Tends to be lost in second phase, step in scoping phase
Informal networks become less significant, relying more and more on data on screen
Characterization as network doesn’t work
Implicit processing system
Traders like skiing
Being a physical activity, like sexual assault when the market turns against the trader
Situated learning could be important
Embodied awareness present in the practice
Question about algorithms, which has become important to trading practice
Can we attribute agency to trading algorithms?
Caution against treating algorithms as agents
What kind of attributions do traders make about algorithms, and the fantasies they make, leading to performative effects?
Exploration of hybrid people:
People who have an understanding of algorithm, and of market
Speed was important, now it’s the hybridity, insight into the design of the algorithms
Competitive advantage in recruiting and developing people with the hybrid capabilities
People who develop and acquire hybrid capabilities can move away from banks, to independence
Changes structure of the market
Legal context:
Regulation of activities
Issues of corporate governance and public accountability
Future research agenda on regulatory effects, where regulators are not capable of catching up
[Questions]
Bad day for market analysts. Philosophical changes. Are we controlled by algorithms?
Spoke to a metaperson about how to debug algorithms? They change them all of the time, they don’t have time to really debug them?
Choice of wording to catch phenomena? What metapersons?
Do algorithms have emotions? They have rituals? Flash crashes
Collective agency?
Systems drift?
Flash crashes, no shadow?
[Responses by Karin]
Being controlled by algorithms
In some areas, could be worse than in financial markets
e.g. what will self-driving cars do to the male ego?
Maybe not control, but a shift that could have identity consequences
Economists admit it’s an open question
Automation could be different from the steam engine
If we don’t create new jobs, could create new social problems
Consequences are not discussed
In financial market, needs to be control of the algorithms
Flash crashes don’t give up control, they do suggest excess trading that human beings can’t handle, but the financial system does
Meta-persons
No time for debugging
Answer may be short
Firms using algorithms have an interest in debugging, because there could be immediate costs
Could be interested in constructing algorithms so that they don’t do too much damage in one go
Regulators: sane, or cognitively captured?
Regulating from the outside: could create more problems
Luhmann: I touch it, but I touch it with a lot of care
10 years of changes in language:
Trying to stay with usual languages or professionalization doesn’t capture fantasies
After studying physicists and molecular biologists, thought financial markets should be simpler, but it’s not true
The epistemology of markets isn’t the same as epistemology of natural sciences
Rating agencies and analysts are paid by institutions, which creates contamination: Chinese walls don’t really work
Economics and financial progress works
Different language for meta-persons
In native languages, spirits are part of the world
Polity in tribes that don’t have a governance system
A market is supposed to egalitarian, but now has a governance structure
Call it a metapersons, not CEOs
Can’t use hierarchies and markets
Trying to incorporate phenomenology
Do algorithms have emotions?
No
It’s not the algorithms that have rituals
A ritual has solidarity consequences, in the social sciences (Durkheim, Goffman, interaction-ritual-chains)
Not really changing, it’s a symbolic consequence
Increased solidarity, increased identification
Question as sociologist: do we need to integrate algorithms into our sociological notions?
Algorithms bring markets down, but then don’t bring the market down
Governments have dignified the algorithms
Collective agency could be a topic of investigation (not done here)
They’re going to take our jobs, in professionals areas (law, teaching)
Will be a collective effect of algorithms that will affect the social world
Do algorithms shift?
There are learning algorithms
They learn not by what they’re supposed to learn, they respond to the data
A recent development
Flash crashes not financially relevant, but socially relevant
Not a real ritual, would be symbolic
Market is brought down, not just a symbol
Social effects of making algorithms more acceptable