Kevin Slavin @slavin_fpo( at NEXT11 talks about information technology.
Chaptter I: The city as a motherboard
In the 1920ies ppl inveted aitplane listening technology, as planes tended to bring danger. Radar followed and indangered planes in the air - to be sgot down.
An invisible plane bevomes visible once it crashes.
You cannot hide a big plane, but if you make the big plane to look like many small things (like birds) you hided it. But if you scan for 'something electrically powered move through the sky, you most probably are seeing a plane.
THis way you break x years of reserch just by chaning the approach.
You can analyse the financial martket the same way.
Banks are byung i million shares from one source.
Look at small trades adding up to a large trade.
Algorithms analyse and predict.
The speed of the trade matters. Which is why computers are good at this.
A being millisecond faster helps.
Some places are closer to the internet backbone than others, they offer speed advantages.
This is why ypou need to be physically close to the wires, but it is half a mile to wallstreet and that creates tens of million dollar damage. And the real estate near to network hubs starts to spike.
People are moved out of thiose areas, enforce the floors, put servers there.
The market is nowhere and somewhere specific at the same time.
The real estate is not valuable because of what humas do but because of what servers do.
We optimize Manhattan like optimizing a motherboard. For transaction speed. Humans are just loitering there and being ineffective.
Chapter II
Radio playlists are just automated tied to music sales in an area.
Algorithms making decisions are everywhere. Not only in Finance.
The only button is the stop button.
Aven arrests are determined that vway.
THree problems:
- opacity
- inscruablity
- somezhing harder to describe
Opacity
- Elevator with no number buttons on the inside
- the only button left on the inside is 'stop'
- . everything else is determined by the algorithm
- 2 vaccum robots have totally different approaches
-one cleans like a human systematically
- the other seems to move randomly and thus irritatingly
Inscrutablity
- a genetic algorithm
- to create a car
- maybe there is something better than common sense (put wheels on the bottom)
- algorithms don_'t speak in 'hman'
- http://areacodeinc.com/
Corewar
- preogramierduell
- durch algorithmen
- the patterns are similar to stock maret algorithms
- the roots: fances Galton (Darwins cousin)
- ?_ how will cerrtain seeds grow / Prediction
- make something that does not look like data (plats) to data and back to plants.
- cf the work of Oren AShenfelter
- PREDICT THE quality of wine, something other pl do just with the sensorics of their mouth
- the opinions are embedded in the mathematics
Netflix Cinematch
- you might like this
- error margin: ca 1 star out of 5 (.95)
- people created alternative algorithms, improved it by about 10%
The human brain is not a great database.
Algorithms deal with the mess in the brain that is not algorithmic.
What if we knew how good the movie was before it is made?
- Epagogix can alnalyze the script of the moovie and determine what it will earn or if it should be made.
- in this scenario: who is the user?
- Something is optimal tll it is disrupted
- sometimes you do not know why the stock market loses lots of value in a short time. and no human knows why.
(this is when you hit the red button and turn the system off)
How will such a "flashrash" look in datinbg, criminal justice, radio playlists, cinematch, ... would we know that the flashcrash happened?
We tell ourselve stories to make sense of the world.
This is the human weapon.
Chaptter I: The city as a motherboard
In the 1920ies ppl inveted aitplane listening technology, as planes tended to bring danger. Radar followed and indangered planes in the air - to be sgot down.
An invisible plane bevomes visible once it crashes.
You cannot hide a big plane, but if you make the big plane to look like many small things (like birds) you hided it. But if you scan for 'something electrically powered move through the sky, you most probably are seeing a plane.
THis way you break x years of reserch just by chaning the approach.
You can analyse the financial martket the same way.
Banks are byung i million shares from one source.
Look at small trades adding up to a large trade.
Algorithms analyse and predict.
The speed of the trade matters. Which is why computers are good at this.
A being millisecond faster helps.
Some places are closer to the internet backbone than others, they offer speed advantages.
This is why ypou need to be physically close to the wires, but it is half a mile to wallstreet and that creates tens of million dollar damage. And the real estate near to network hubs starts to spike.
People are moved out of thiose areas, enforce the floors, put servers there.
The market is nowhere and somewhere specific at the same time.
The real estate is not valuable because of what humas do but because of what servers do.
We optimize Manhattan like optimizing a motherboard. For transaction speed. Humans are just loitering there and being ineffective.
Chapter II
Radio playlists are just automated tied to music sales in an area.
Algorithms making decisions are everywhere. Not only in Finance.
The only button is the stop button.
Aven arrests are determined that vway.
THree problems:
- opacity
- inscruablity
- somezhing harder to describe
Opacity
- Elevator with no number buttons on the inside
- the only button left on the inside is 'stop'
- . everything else is determined by the algorithm
- 2 vaccum robots have totally different approaches
-one cleans like a human systematically
- the other seems to move randomly and thus irritatingly
Inscrutablity
- a genetic algorithm
- to create a car
- maybe there is something better than common sense (put wheels on the bottom)
- algorithms don_'t speak in 'hman'
- http://areacodeinc.com/
Corewar
- preogramierduell
- durch algorithmen
- the patterns are similar to stock maret algorithms
- the roots: fances Galton (Darwins cousin)
- ?_ how will cerrtain seeds grow / Prediction
- make something that does not look like data (plats) to data and back to plants.
- cf the work of Oren AShenfelter
- PREDICT THE quality of wine, something other pl do just with the sensorics of their mouth
- the opinions are embedded in the mathematics
Netflix Cinematch
- you might like this
- error margin: ca 1 star out of 5 (.95)
- people created alternative algorithms, improved it by about 10%
The human brain is not a great database.
Algorithms deal with the mess in the brain that is not algorithmic.
What if we knew how good the movie was before it is made?
- Epagogix can alnalyze the script of the moovie and determine what it will earn or if it should be made.
- in this scenario: who is the user?
- Something is optimal tll it is disrupted
- sometimes you do not know why the stock market loses lots of value in a short time. and no human knows why.
(this is when you hit the red button and turn the system off)
How will such a "flashrash" look in datinbg, criminal justice, radio playlists, cinematch, ... would we know that the flashcrash happened?
We tell ourselve stories to make sense of the world.
This is the human weapon.
Kommentare
Kommentar veröffentlichen