CS 3043 Social Implications Of Computing Class 10

CS 3043 Social Implications Of Computing Class 10 Errors Failures Risks Keith A. Pray Instructor socialimps. keithpray. net 11/1/2020 © 2013 Keith A. Pray 1

CS 3043 Social Implications Of Computing Overview 1. 2. 3. 4. Errors, Failures, And Risks Assignment Students Present Guest Speaker 11/1/2020 © 2013 Keith A. Pray 2

CS 3043 Social Implications Of Computing http: //pbfcomics. com/197/ 11/1/2020 © 2012 Keith A. Pray 3

CS 3043 Social Implications Of Computing Overview 1. 2. 3. 4. Errors, Failures, And Risks Assignment Students Present Guest Speaker 11/1/2020 © 2013 Keith A. Pray 4

CS 3043 Social Implications Of Computing Assignment n Finish reading n First draft of group presentation Due Friday 2: 59 PM ¨ Post on group website ¨ Email presentation to me and TA ¨ 11/1/2020 © 2013 Keith A. Pray 5

CS 3043 Social Implications Of Computing Overview 1. 2. 3. 4. Errors, Failures, And Risks Assignment Students Present Guest Speaker 11/1/2020 © 2013 Keith A. Pray 6

CS 3043 Social Implications Of Computing Madalyn Coryea Did it work? The Importance of Feedback and User Frustration n Feedback assures users ¨ n Computers appear unreliable ¨ n Reinforces that their task is completed correctly [6] Lack of feedback may be interpreted as no effect [4] Frustrate, confuse, or cause users to make mistakes [6] No feedback present, no error seen ¨ Users become unable to rectify errors ¨ [4] Jakob Nielson, “Error Message Guidelines” [6] Manuel A. Pérez-Quiñones and John L. Sibert, “A Collaborative Model of Feedback in Human-Computer Interaction” 11/1/2020 © 2013 Keith A. Pray 7

CS 3043 Social Implications Of Computing What is feedback? n Definition ¨ n Madalyn Coryea “Communication from the system to the user as a direct result of a user's action” [6] Importance Provides evidence of closure [6] ¨ Communication expectations [3][6] ¨ Confirms successful steps in completion of goal [2][6] ¨ n Types Error messages ¨ Clicking sound to hover menus ¨ [2] Jonathan Lazar et al. “Severity and Impact of Computer User Frustration: A Comparison of Student and Workplace Users” [3] B. Jean Mason and Roger Bruning, “Providing Feedback in Computer-based Instruction: What the Research Tells Us ” [6] Manuel A. Pérez-Quiñones and John L. Sibert, “A Collaborative Model of Feedback in Human-Computer Interaction” 11/1/2020 © 2013 Keith A. Pray 8

CS 3043 Social Implications Of Computing Madalyn Coryea User Frustration: Causes & Effects n Direct Causes ¨ n Lost time, lost work, inability to reach goal [2][7] Indirect Causes Poor usability [2] ¨ Poor feedback/Error messages [2][7] ¨ n How much does Frustration affect us? 33 -50% of time spent on computer wasted due to frustration [2][7] ¨ 80% of users cursed at computer [2] ¨ [2] Jonathan Lazar et al. “Severity and Impact of Computer User Frustration: A Comparison of Student and Workplace Users” [7] ] Ceaparu, I. , Lazar et al. (2004). “Determining Causes and Severity of End-User Frustration” 11/1/2020 © 2013 Keith A. Pray 9

CS 3043 Social Implications Of Computing User-Frustration Study, 2004 n Madalyn Coryea Survey Conditions [2] 107 Students (aged 18 -22) ¨ one-hour, computer task of their choosing ¨ n Results [2] 357 frustrating experiences ¨ Most frustrating tasks: ¨ Web-browsing (120), email (50), word processing (44) ¨ Primary cause: error messages (35 cases) [2] Jonathan Lazar et al. “Severity and Impact of Computer User Frustration: A Comparison of Student and Workplace Users” 11/1/2020 © 2013 Keith A. Pray 10

CS 3043 Social Implications Of Computing Error Messages, dos & don’ts n Madalyn Coryea Oracle’s guide for error messages [5] What happened ¨ Why it happened ¨ What should be done to correct the problem ¨ n Figure One: A poor error message Other guidelines [4] Neilson Norman Group guidelines ¨ Explicit, Human-Readable, Polite, Precise, Constructive Advice ¨ [4] Jakob Nielson, “Error Message Guidelines” [5] Oracle, “Guidelines for Error Messages” 11/1/2020 © 2013 Keith A. Pray Figure Two: A good error message according to Oracle’s standards 11

CS 3043 Social Implications Of Computing Madalyn Coryea Lack of User Feedback n “The very worst error messages are those that don't exist. ” [4] ¨ n Data is misinterpreted because of improper or lack of feedback Hamilton, Ontario, July 1985 [1] “According to the display, the linear accelerator had not yet delivered any radiation to the patient. ” [1] ¨ 65 -85 times normal dose of radiation [1] ¨ [1] Quinn, Michael J. , Ethics for the Information Age. 5 th ed. , [4] Jakob Nielson, “Error Message Guidelines” 11/1/2020 © 2013 Keith A. Pray 12

CS 3043 Social Implications Of Computing Feedback is not an afterthought n Madalyn Coryea Feedback should be part of initial design Walk user through [3] ¨ May never put it in ¨ MVC is good ¨ n More than Error Messages User frustration [2] ¨ Fix it yourself ¨ Every action needs reaction ¨ n Figure Three: Firefox’s classic “Well, this is embarrassing” Software poorly designed, not worth using ¨ Software is for humans [2] Jonathan Lazar et al. “Severity and Impact of Computer User Frustration: A Comparison of Student and Workplace Users ” [3] B. Jean Mason and Roger Bruning, “Providing Feedback in Computer-based Instruction: What the Research Tells Us” 13 © 2013 Keith A. Pray 11/1/2020

CS 3043 Social Implications Of Computing Sources n n n n n Madalyn Coryea [1] Quinn, Michael J. , Ethics for the Information Age. 5 th ed. , New Jersey: Pearson, 2005, pg. 372 [2] Jonathan Lazar, Adam Jones, Mary Hackley, and Ben Shneiderman, “Severity and Impact of Computer User Frustration: A Comparison of Student and Workplace Users”, December 23, 2004, http: //www. cs. umd. edu/~ben/papers/Lazar 2006 Severity. pdf (Accessed 9/29/13) [3] B. Jean Mason and Roger Bruning, “Providing Feedback in Computer-based Instruction: What the Research Tells Us”, http: //dwb. unl. edu/Edit/MB/Mason. Bruning. html (Accessed 9/29/13) [4] Jakob Nielson, “Error Message Guidelines”, 6/24/01, http: //www. nngroup. com/articles/errormessage-guidelines/ (Accessed 9/29/13) [5] Oracle, “Guidelines for Error Messages”, 2010 http: //docs. oracle. com/cd/E 19683 -01/8160284/sgmessages-3/index. html (Accessed 9/29/13) [6] Manuel A. Pérez-Quiñones and John L. Sibert, “A Collaborative Model of Feedback in Human. Computer Interaction”, http: //www. sigchi. org/chi 96/proceedings/papers/Perez/map 1 txt. htm (Accessed 9/29/13) [7] Ceaparu, I. , Lazar, J. , Bessiere, K. , Robinson, J. , and Shneiderman, B. (2004). Determining Causes and Severity of End-User Frustration. International Journal of Human-Computer Interaction, 17 (3), 333 -356 Figure 1: “Java Web Start installation and troubleshooting”, http: //bioinformatics. cmuj. krakow. pl/reveal/cgi-bin/problem. pl Figure 2: “Guidelines for Error Messages” 2010 http: //docs. oracle. com/cd/E 19683 -01/8160284/sgmessages-3/index. html 11/1/2020 © 2013 Keith A. Pray 14

CS 3043 Social Implications Of Computing Sources n Figure 3: “Well, this is embarrassing”, 2010, http: //www. codinguser. com/2010/07/well-this-isembarrassing/ (Accessed 10/1/13) 11/1/2020 © 2013 Keith A. Pray 15

CS 3043 Social Implications Of Computing Yilan Liu Oops! Our Website is Down n n Performance practice at major companies are often considered good, though not perfect [1] Even for companies with best practice, outages are sometimes unavoidable 11/1/2020 © 2013 Keith A. Pray 16

CS 3043 Social Implications Of Computing Yilan Liu Case Study – Google 08/16/13 n n The official downtime was between 1 and 5 minutes [2] Estimated lost $500, 000 in revenue [3] Internet traffic dropped by 40% globally [2] Google barely released any information about the cause of this outage [4] 11/1/2020 © 2013 Keith A. Pray 17

CS 3043 Social Implications Of Computing Yilan Liu Case Study – Google 08/16/13 11/1/2020 © 2013 Keith A. Pray 18

CS 3043 Social Implications Of Computing Yilan Liu Case Study – Amazon 12/24/12 n n n Elastic Load Balancing Service of Amazon Web Services (AWS) experienced an 23 -hour-long outage in its US-East-1 region [5] Took down other web-based services completely, mostly notably Netflix [6] Netflix referred to this as “bad timing” [6] 11/1/2020 © 2013 Keith A. Pray 19

CS 3043 Social Implications Of Computing Yilan Liu Case Study – Amazon 12/24/12 n n n Fear has risen against the reliability of not only AWS, but cloud services in general [7] Netflix was considering isolating and migrating its service [8] Amazon released detailed summary of this outage, blaming human error instead of Christmas Eve traffic [5] 11/1/2020 © 2013 Keith A. Pray 20

CS 3043 Social Implications Of Computing Yilan Liu Case Study – Linked. In 06/19/13 n n n Down for 6. 5 hours [9] Users were redirected, or couldn’t reach the site at all [9] Observed failures on other websites such as The Atlantic, The Daily Beast, and The Economist. [10] 11/1/2020 © 2013 Keith A. Pray 21

CS 3043 Social Implications Of Computing Yilan Liu Case Study – Linked. In 06/19/13 n n n Caused by human error at Network Solutions, provider for Linked. In’s DNS [9] Impacted sites referenced an inline Java. Script call to “platform. linkedin. com” - Single Point of Failure [10] Great transparency shown by both Linked. In and Network Solutions 11/1/2020 © 2013 Keith A. Pray 22

CS 3043 Social Implications Of Computing Yilan Liu Case Study – Linked. In 06/19/13 11/1/2020 © 2013 Keith A. Pray 23

CS 3043 Social Implications Of Computing Yilan Liu Impact of Outages Upset users, loss of their productivity n Revenue loss n Brand name damage n Weaken trust in the company and the industry n The Domino Effect n Panic in public [11] n 11/1/2020 © 2013 Keith A. Pray 24

CS 3043 Social Implications Of Computing Yilan Liu Conclusion - Cause of Outages n n Many would think the outages are caused by bugs - insufficient testing, defects in code However, in many cases the actual cause is: Human error – improper training and loose control o Infrastructures that are too intertwined and dependent on each other [7] o Unavoidable cause like weather [12] o 11/1/2020 © 2013 Keith A. Pray 25

CS 3043 Social Implications Of Computing Yilan Liu Conclusion - What can we do? n Reduce the chance of outages: Proper training and strict control to avoid human errors o Isolate and distribute services and infrastructure to avoid outages o n Redemption after outages: Automatic roll-back [7] o Faster emergency response system o 11/1/2020 © 2013 Keith A. Pray 26

CS 3043 Social Implications Of Computing Yilan Liu Sources 1 Barber, Scott. "Any Given Monday – Google, Microsoft and Amazon All Experience Outages. " Giga. OM. N. p. , 20 Aug. 2013. Web. 30 Sept. 2013. <http: //blog. smartbear. com/performance-testing/any-given-monday-google-microsoft-and-amazon-allexperience-outages/>. 2 Mack, Eric. "Google Outage Reportedly Caused Big Drop in Global Traffic. " CNET News. CBS Interactive, 16 Aug. 2013. Web. 30 Sept. 2013. <http: //news. cnet. com/8301 -17938_105 -57598978 -1/google-outage-reportedly-caused-big-drop-in-global-traffic/>. 3 "Google Outage: Internet Traffic Plunges 40%. " Sky. com. N. p. , 17 Aug. 2013. Web. 30 Sept. 2013. <http: //news. sky. com/story/1129847/google-outage-internet-traffic-plunges-40 -percent>. 4 Nagy, Mandy. "Why Is Google Silent on Its Outage? " Legal Insurrection. N. p. , 19 Aug. 2013. Web. 30 Sept. 2013. <http: //legalinsurrection. com/2013/08/why-is-google-silent-on-its-outage/>. 5 "Summary of the December 24, 2012 Amazon ELB Service Event in the US-East Region. " AWS. Amazon. com, 25 Dec. 2012. Web. 30 Sept. 2013. <http: //aws. amazon. com/cn/message/680587/>. 6 Harris, Derrick. “Amazon Blames Human Error for Xmas Eve Outage; Netflix Vows Better Resiliency — Tech News and Analysis. ” Giga. OM. N. p. , 31 Dec. 2012. Web. 30 Sept. 2013. <http: //gigaom. com/2012/12/31/amazon-blames-human-error-forxmas-eve-outage-netflix-vows-better-resiliency/>. 11/1/2020 © 2013 Keith A. Pray 27

CS 3043 Social Implications Of Computing Yilan Liu Sources 7 Williams, Wayne. "UC 4: Web Outages, like Amazon's Recent Issues, Can Be Predicted and Avoided [Q&A]. " Beta. News. N. p. , 04 Jan. 2013. Web. 30 Sept. 2013. <http: //betanews. com/2013/01/04/uc 4 -web-outages-like-amazons-recent-issues-can-bepredicted-and-avoided-qa/>. 8 Cockcroft, Adrian. "The Netflix Tech Blog: A Closer Look At The Christmas Eve Outage. " The Netflix Tech Blog. N. p. , 31 Dec. 2012. Web. 30 Sept. 2013. <http: //techblog. netflix. com/2012/12/a-closer-look-at-christmas-eve-outage. html>. 9 Fisher, Dennis. "LINKEDIN OUTAGE TIED TO DNS ISSUE. " Threatpost. N. p. , 20 June 2013. Web. 30 Sept. 2013. <http: //threatpost. com/linkedin-outage-tied-to-dns-issue>. 10 "The Domino Effect of Linked. In’s DNS Outage. " Catchpoint. N. p. , 21 June 2013. Web. 30 Sept. 2013. <http: //blog. catchpoint. com/2013/06/21/the-domino-effect-of-linkedin-dns-outage/>. 11 Underwood, Madison. "A Short Amazon Website Outage Caused Twitter to Freak out. "Al. com. N. p. , 19 Aug. 2013. Web. 30 Sept. 2013. <http: //blog. al. com/wire/2013/08/amazons_website_goes_down_and. html>. 12 Miller, Rich. "Multiple Generator Failures Caused Amazon Outage. " Data Center Knowledge. N. p. , 03 July 2012. Web. 01 Oct. 2013. <http: //www. datacenterknowledge. com/archives/2012/07/03/multiple-generator-failures-caused-amazon-outage/>. 11/1/2020 © 2013 Keith A. Pray 28

CS 3043 Social Implications Of Computing Yilan Liu Image Sources P 3 Mack, Eric. "Google Outage Reportedly Caused Big Drop in Global Traffic. " CNET News. CBS Interactive, 16 Aug. 2013. Web. 30 Sept. 2013. <http: //news. cnet. com/8301 -17938_105 -57598978 -1/google-outage-reportedly-caused-big-drop-in-global-traffic/>. P 8 "The Domino Effect of Linked. In’s DNS Outage. " Catchpoint. N. p. , 21 June 2013. Web. 30 Sept. 2013. <http: //blog. catchpoint. com/2013/06/21/the-domino-effect-of-linkedin-dns-outage/>. P 9 Underwood, Madison. “A Short Amazon Website Outage Caused Twitter to Freak out. ”Al. com. N. p. , 19 Aug. 2013. Web. 30 Sept. 2013. <http: //blog. al. com/wire/2013/08/amazons_website_goes_down_and. html>. 11/1/2020 © 2013 Keith A. Pray 29

CS 3043 Social Implications Of Computing Aquame Matthews Medical Databases n n Lack of standards Data Entry Referrals, Charges, and Prescriptions ¨ 16. 5% of medication errors [3] ¨ Case of wrong dosage at a county jail [1]. ¨ 11/1/2020 © 2013 Keith A. Pray 30

15. 00% 10. 00% 5. 00% 0. 00% Pr rm oc an ed ce Tr ure de f an s sc /pro icit rip t tio oco l C n D om ina not oc c f um put cur ollo en er e ate we / n t d K no atio try omi tte n w le in d dg ad Co e e m def qua m un icit te o r l i D ac ru Wri cati ki o g tte n ng d n D ist o isp rib rd en ut er sin ion g de sys In Ab vic tem co e b rre rev pr pro Sy ct iat obl ble ste en ion em m s m try sa b y Ca fegu pre Ill lcu ard scr eg i l ib atio s in ber In le o n e ad q r eq r un ror uat e o D uat cle os a r l ag e m r h ac ki Fa e fo oni and ng t w x/ o sc rm c rin rit in an g g ne onfu r p s ro ion bl em s rfo Pe CS 3043 Social Implications Of Computing 40. 00% 35. 00% 30. 00% 25. 00% 20. 00% [3] 11/1/2020 © 2013 Keith A. Pray 31

CS 3043 Social Implications Of Computing Aquame Matthews Issues (cont. ) n Security Hacked for health insurance data, personal and electronic payment transaction data [6] ¨ Australian medical center held ransom for demand of $4, 000 AUD [7]. ¨ n System Failure ¨ n One case at the University of Pittsburgh Medical Center, parts of their system were down for six hours over a two day period [1]. Rip and Replace ¨ 11/1/2020 Can take years to complete [5] © 2013 Keith A. Pray 32

CS 3043 Social Implications Of Computing Aquame Matthews Trak. Care n World’s leading Internet-based healthcare information system [4]. n Main Features: Patient record centric ¨ Can easily be tailored to fit information requirements and workflow preferences ¨ n Built on Ensemble 11/1/2020 © 2013 Keith A. Pray 33

CS 3043 Social Implications Of Computing Aquame Matthews Trak. Care (cont. ) [4] 11/1/2020 © 2013 Keith A. Pray 34

CS 3043 Social Implications Of Computing [4] 11/1/2020 © 2013 Keith A. Pray 35

CS 3043 Social Implications Of Computing Make the switch! n n n EPR centric models are the way to go. Resolves the issue of standards Inter. Systems addresses rip and replace 11/1/2020 © 2013 Keith A. Pray 36

CS 3043 Social Implications Of Computing Aquame Matthews Sources n n n n 1. MILT FREUDENHEIM, The Ups and Downs of Electronic Medical Records, http: //www. nytimes. com/2012/10/09/health/the-ups-and-downs-of-electronic-medical-records-the-digitaldoctor. html? pagewanted=all&_r=0 (9/25/13) 2. William R. Hersh , The Electronic Medical Record: Promises and Problems, http: //skynet. ohsu. edu/~hersh/jasis-95 -emr. pdf (9/25/13) 3. ANDIS ROBEZNIEKS, Data entry is a top cause of medication errors, http: //www. amednews. com/article/20050124/profession/301249959/4/ (9/25/13) 4. Inter. Systems, Internet-based Healthcare Information System, http: //www. intersystems. com/trakcare/features/overview. html (9/25/13) 5. Inter. Systems, 2. 2 Trak. Care Foundation CIS (Clinical Information System), http: //www. intersystems. com/trakcare/solution/section-02_2. html (9/25/13) 6. NTC, Health Care Data Breaches Target Point-of-Sale. , http: //www. nationaltransaction. com/credit-cardmerchant/health-care-data-breaches-target-point-of-sale/#. Ukp. FVIZ 02 So (9/30/13) 7. John E. Dunn, Ransom hackers encrypt medical centre's entire database http: //news. techworld. com/security/3415635/ransom-hackers-encrypt-medical-centres-entire-database/ (9/30/13) 11/1/2020 © 2013 Keith A. Pray 37

CS 3043 Social Implications Of Computing Overview 1. 2. 3. 4. Security & Errors, Failures, And Risks Assignment Students Present Guest Speaker 11/1/2020 © 2013 Keith A. Pray 38

CS 3043 Social Implications Of Computing Class 10 The End Keith A. Pray Instructor socialimps. keithpray. net 11/1/2020 © 2013 Keith A. Pray 39

CS 3043 Social Implications Of Computing Risk n n What is it? Chance of it happening? ¨ n Random, Historical statistics, Analysis Give examples of risks with computer systems 11/1/2020 © 2013 Keith A. Pray 40

CS 3043 Social Implications Of Computing Predicting Risk n n n What are the factors? Enough data to be statistically significant? Can you reduce or increase your own risk? Chance ¨ Severity ¨ 11/1/2020 © 2013 Keith A. Pray 41

CS 3043 Social Implications Of Computing Reliability Of Statistics n n n Are other factors controlled? Is enough time covered? Is all data reported? 11/1/2020 © 2013 Keith A. Pray 42

CS 3043 Social Implications Of Computing Cost-Benefit Analysis n n How much does it cost? What do I get for this cost? 11/1/2020 © 2013 Keith A. Pray 43

CS 3043 Social Implications Of Computing Risk-Benefit Analysis n n n What could go wrong? How likely is it? How costly are the consequences? 11/1/2020 © 2013 Keith A. Pray 44

CS 3043 Social Implications Of Computing Limitations to Risk-Benefit Analysis n n Hard to quantify probabilities Hard to quantify costs Who bears the costs? Are the potential benefits worth it? 11/1/2020 © 2013 Keith A. Pray 45

CS 3043 Social Implications Of Computing Some Measures n n n Mean Time To Failure (MTTF) Mean Time Between Failures (MTBF) Mean Time To Repair (MTTR) 11/1/2020 © 2013 Keith A. Pray 46

CS 3043 Social Implications Of Computing Relying Too Much n n Limits of modeling reality Limits of precision Limits of algorithms Limits of interpretation 11/1/2020 © 2013 Keith A. Pray 47

CS 3043 Social Implications Of Computing Value of Intuition n Are results reasonable? 11/1/2020 © 2013 Keith A. Pray 48

CS 3043 Social Implications Of Computing Producing Good Software n n n Good specifications Good usability Good design Robustness Good implementation Testing 11/1/2020 © 2013 Keith A. Pray 49

CS 3043 Social Implications Of Computing Plan For The Long Term n n n n Version control Future development New platforms Patches Security Training Customer support 11/1/2020 © 2013 Keith A. Pray 50