Trent McConaghy is an AI researcher and blockchain engineer. He is the Founder & CTO of BigchainDB. He started doing AI research for national defense as an undergrad, going on to obtain a PhD from KU Leuven. He has done Machine Learning research for the Canadian Department of National Defense, and has written two books and 35 papers, and holds 20 patents on Machine Learning, circuits and creativity.
8. “GOOG and FB now have direct
influence over 70%+ of internet
traffic.”
In “The Web Began Dying in 2014”
by Andrew Staltz
https://staltz.com/the-web-began-dying-in-2014-heres-how.html
13. Silo Pool Mo’ Data
(and mo’ compute)
Mo’ accuracy
Mo’ $
Have lotsa data
(1000
enterprises)
Have lotsa AI
(1000 AI
startups)
14. Silo Pool Mo’ Data
(and mo’ compute)
Mo’ accuracy
Mo’ $
Have lotsa AI
(1000 AI
startups)
Have lotsa data
(1000
enterprises)
15. Silo Pool Mo’ Data
(and mo’ compute)
Mo’ accuracy
Mo’ $
Have lotsa AI
(1000 AI
startups)
Data
marketplace
DM DM DM
DM DM DM DM
Have lotsa data
(1000
enterprises)
16. Silo Pool Mo’ Data
(and mo’ compute)
Mo’ accuracy
Mo’ $
Change the status quo?
Have lotsa AI
(1000 AI
startups)
Data
marketplace
DM DM DM
DM DM DM DM
Ocean = universal data economy
Curated data in a decentralized substrate
Have lotsa data
(1000
enterprises)
19. What’s the amazing thing about
blockchains?
•Decentralized?
•Immutability?
•Aligning incentives?
•Raising $$$$ off a whitepaper and a dog?
•One more…
20. The Not-So-Obvious Blockchain Superpower:
You can get people to do things
So, what do you want people to do?
Design economic incentives for that thing,
and bake it into the network as block rewards
21. Economic Incentive for Bitcoin
Objective: Maximize security of chain
• Where “security” = compute power
• Therefore, super expensive to roll back changes to the transaction log
E(Ri) α Hi * T
• E() = expected value
• Ri = block rewards
• Hi = hash power of actor
• T = # tokens (BTC) dispensed each block
22. Bitcoin Mining Rewards, from Economic Incentive
You get block rewards if:
•You’re running a full node
•And you have gobs of compute power
24. Economic Incentive for Ocean
Objective: Maximize supply of high-quality (curated) data
• This means: reward curating data + making it available
• Where “curating” = betting on data. Reward taste-making.
E(Rij) α log10(Sij) * log10(Dj) * T *Ri
• E(Rij) = expected block rewards
• Sij = stake in dataset j
• Dj = popularity = # downloads
• T = # tokens during interval
• Ri = download ratio (prevents some attacks)
25. Ocean Mining Rewards, from Economic
Incentive
You get block rewards if:
• You’ve bet on a dataset (i.e. curated)
• And you make it available when asked
26. Ocean Blockchain Network Ocean core software –
Miner 1
Block rewards
mechanism
Verify
data is
made
available
Data
marketplace
DM DM DM DM
Miner
Miner
Miner
Miner
Miner
Curation market
Actors registry
29. Equalize the
opportunity for
access to data
and
benefits of AI
Have lotsa AI
(1000 AI
startups)
Data
marketplace
DM DM DM
DM DM DM DM
Have lotsa data
(1000
enterprises)
Universal data economy
34. We’ve lost the web.
Let’s reclaim it, better than ever...
Trent McConaghy
@trentmc0
Let’s build a new data economy
that gives power to the people!
Editor's Notes
Despite successes here and there, up until recently AI was not widespread. For decades, AI researchers were churning out papers like crazy with all their new pet algorithms, on the same shared datasets; but there was often barely a dent in accuracy. You’d go from 25% to 24% error, still super far away from something usable which usually needs at least 1%.
But then a breakthrough occurred. Researchers started to realize that maybe, just maybe, more *data* could help.
This plot shows how with the same algorithm, if you increase the size of the dataset by 1000x, you can bring error from 25% to 3%. The choice of the algorithm hardly mattered in fact. What matters was the amount of data. Data was unreasonably effective.
Researchers at Google called this “the unreasonable effectiveness of data” and published a paper on it in 2007. Since then, their driving mission has been “get more data”.
This “data” realization happened in 2005. A decade later, some consider “data” to be the new oil, to be the “world’s most valuable resource”. And that’s not far off.
More data (and the ability to compute on it) leads to more accurate models, leads to more money.
AI has taken off in the last five years because the UX for doing AI has improved. Now you just need massive data and compute, and you can have a fighting chance at solving problems that were previously intractable.
Algorithms are table stakes. Data is the moat. This is why Google is giving away their algorithms and their processing chips. They buy satellite companies just for the data. And in case you hadn’t noticed, they’re not giving it away.
This “data” realization happened in 2005. A decade later, some consider “data” to be the new oil, to be the “world’s most valuable resource”. And that’s not far off.
More data (and the ability to compute on it) leads to more accurate models, leads to more money.
AI has taken off in the last five years because the UX for doing AI has improved. Now you just need massive data and compute, and you can have a fighting chance at solving problems that were previously intractable.
Algorithms are table stakes. Data is the moat. This is why Google is giving away their algorithms and their processing chips. They buy satellite companies just for the data. And in case you hadn’t noticed, they’re not giving it away.
This “data” realization happened in 2005. A decade later, some consider “data” to be the new oil, to be the “world’s most valuable resource”. And that’s not far off.
More data (and the ability to compute on it) leads to more accurate models, leads to more money.
AI has taken off in the last five years because the UX for doing AI has improved. Now you just need massive data and compute, and you can have a fighting chance at solving problems that were previously intractable.
Algorithms are table stakes. Data is the moat. This is why Google is giving away their algorithms and their processing chips. They buy satellite companies just for the data. And in case you hadn’t noticed, they’re not giving it away.
This “data” realization happened in 2005. A decade later, some consider “data” to be the new oil, to be the “world’s most valuable resource”. And that’s not far off.
More data (and the ability to compute on it) leads to more accurate models, leads to more money.
AI has taken off in the last five years because the UX for doing AI has improved. Now you just need massive data and compute, and you can have a fighting chance at solving problems that were previously intractable.
Algorithms are table stakes. Data is the moat. This is why Google is giving away their algorithms and their processing chips. They buy satellite companies just for the data. And in case you hadn’t noticed, they’re not giving it away.
This “data” realization happened in 2005. A decade later, some consider “data” to be the new oil, to be the “world’s most valuable resource”. And that’s not far off.
More data (and the ability to compute on it) leads to more accurate models, leads to more money.
AI has taken off in the last five years because the UX for doing AI has improved. Now you just need massive data and compute, and you can have a fighting chance at solving problems that were previously intractable.
Algorithms are table stakes. Data is the moat. This is why Google is giving away their algorithms and their processing chips. They buy satellite companies just for the data. And in case you hadn’t noticed, they’re not giving it away.
This “data” realization happened in 2005. A decade later, some consider “data” to be the new oil, to be the “world’s most valuable resource”. And that’s not far off.
More data (and the ability to compute on it) leads to more accurate models, leads to more money.
AI has taken off in the last five years because the UX for doing AI has improved. Now you just need massive data and compute, and you can have a fighting chance at solving problems that were previously intractable.
Algorithms are table stakes. Data is the moat. This is why Google is giving away their algorithms and their processing chips. They buy satellite companies just for the data. And in case you hadn’t noticed, they’re not giving it away.
This “data” realization happened in 2005. A decade later, some consider “data” to be the new oil, to be the “world’s most valuable resource”. And that’s not far off.
More data (and the ability to compute on it) leads to more accurate models, leads to more money.
AI has taken off in the last five years because the UX for doing AI has improved. Now you just need massive data and compute, and you can have a fighting chance at solving problems that were previously intractable.
Algorithms are table stakes. Data is the moat. This is why Google is giving away their algorithms and their processing chips. They buy satellite companies just for the data. And in case you hadn’t noticed, they’re not giving it away.