David Woods

Releasing the Adaptive Power of Human Systems. Likes are not endorsements.

[David Woods @ddwoods2]

@emollick Yes, glad you have recognized this fundamental risk for all automata. Plenty of work on this in Resilience Engineering with formal foundations. Gold rush momentum misses these directions

[David Woods @ddwoods2]

@dennismhogan Dose of reality: higher ed is a free (matching) market, choice abounds, distortions occur (legacy) Roth 2012 Nobel. Running Uni as if corporation->distorts ed&research&admin balance w/in & across schools (market pressures too) Affordable access issues

[David Woods @ddwoods2]

New piece on AI limits, safety & testing (joint activity testing) based on large study in 1 of Nature journals https://t.co/ReTjngBSV8

[David Woods @ddwoods2]

@emollick please, decades of work 1 read Behind Human Error 1994/2010 2 u point vaguely in right direction-joint cognitive systems read Woods/Hollnagel 2006. 3 see current work on architecture of layered networks Doyle DeSS or me TGE. I can send u material.

[David Woods @ddwoods2]

here is a new piece, out this morning, on Aviation Safety in the US given recent events. Maybe useful beyond aviation and a reasonable brief explanation of proactive safety (and we never stop needing these).
https://t.co/GSl5xoKHgt

[David Woods @ddwoods2]

@tyler_m_john ground station for autonomous a/c = ground pilot, dog, flight data recorder. Person is there to feed the dog, dog is there to bite person if touches anything, flight data recorder is there to prove when things go wrong-due to human error.

[David Woods @ddwoods2]

@tyler_m_john But why do some say 'swiss cheese' is DinD? B/c actual finding threatens orgs. (multiple contributors, most present in org prior to incident) 97 title=Org. Accidents. I watched in 90's as orgs inverted concepts to appear to adopt still blame events on people

[David Woods @ddwoods2]

@tyler_m_john You have made common error on how complex systems fail & 'swiss cheese' Finding was DinD adds complexities & new vulnerabilities. Linearity fails Defenses erode under production pressure Read Behind Hu. Err. 1994/2010 https://t.co/RSKk1ikmwP

[David Woods @ddwoods2]

@MorePerfectUS Turn a debt trap into debtor prison. Impossible to escape Since can’t declare bankruptcy. Most have already paid much more than the principal. Only the Godfather can admire the theft. The Dems messaging on this is abysmal

[David Woods @ddwoods2]

@emollick 'the objectivity of AI in responding' What? Cook's original came from studies w/ detailed observation of people coping w/ complexity across many settings-a short empirical summary to get people to read our 1994 book Behind Human Error, 2nd ed 2010

[David Woods @ddwoods2]

@emollick As RI Cook's colleague & co-author from 89 to his death, there is context for his 1998 piece & big advances since. Your claims about AI Gold Rush 2 do not match why Cook put original together as a cautionary piece

[David Woods @ddwoods2]

@emollick U realize these are all basics from field of Resiience Engineering (RE)-repeated w/out attribution #22 quotes me in talks/papers; #25 is the basis for RE, theorem (24 years old) & empirical finding (decades old) #23 is 1st line https://t.co/Q6cmd9NXpn ...

[David Woods @ddwoods2]

@tdietterich It’s called context cued retrieval & is basic cognitive psychology. Branford & Franks did modern studies @1970, but much older. Framing/reframing are more context bound effects from being situated

[David Woods @ddwoods2]

@seanluomdphd @MichaelLinLab No. Better ways to increase risk taking in funded projects.

[David Woods @ddwoods2]

@norootcause Lorin, the framing effect is operating here. Our finding is analyze yourself first then use the algorithm for best joint performance

[David Woods @ddwoods2]

@LuizaJarovsky @DanielSolove @HideyukiMATSUMI Abysmal. total ignorance of engineering work for over 40 years (& longer) w/ findings on joint systems (strong & weak) & designs of in many settings. Paper is a hash of weak to worse joint architectures. Human machine eng malpractice

[David Woods @ddwoods2]

@ESYudkowsky Much deeper issue everywhere; See Theory of Graceful Extensibility 2018, Video series on foundations of Resilience Engineering https://t.co/b3GB0snmFz

[David Woods @ddwoods2]

@hendrycks Independence is a critical ingredient in safety; shown over and over. But independence needs to be part of the safety architecture not just an admirable decision by one leader. See how to design safety org 2006 at my research gate page

[David Woods @ddwoods2]

@theseverine Now u see into the root ironies of expertise, tech change, adaption. Contact with variability is needed to handle future change which will vary from past. When tech blocks contact w variability, get deskilling There is way out

[David Woods @ddwoods2]

@ben_golub Reminder: Stories of tech change envision the new forms of congestion cascade & conflict that arise when projected benefits are hijacked How do we know? Empirical studies of coping w complexity Do economists use the results?