usable security- it isn't secure if people can't use it. o-isc conference 14mar2012

39
Usable Security: It isn’t secure if people can’t use it. Darren Kall @darrenkall #secUX

Upload: darren-kall

Post on 07-Nov-2014

330 views

Category:

Technology


0 download

DESCRIPTION

This is one of two talks. This one encourages the security community to adopt a user experience approach to the development and deployment of security products. The second encourages the user experience community to focus their skills on usable security issues. Security products and security issues do not get enough attention from user experience. Yet user experience is at the root cause of many, if not most, security issues. The weakest link in security is not technology but the gap between technology and people. The developer, IT implementer, administrator, and end-user each create vulnerabilities if the system wasn’t designed to be usable for each of them. Technology, policies, management and metrics all improve with a user-centric approach that merges development, security implementation and monitoring with usability.

TRANSCRIPT

Page 1: Usable security- It isn't secure if people can't use it. O-ISC conference 14mar2012

Usable Security:It isn’t secure if people can’t use it.

Darren Kall @darrenkall#secUX

Page 2: Usable security- It isn't secure if people can't use it. O-ISC conference 14mar2012

Employment KALL Consulting Microsoft

◦ Windows Security User Experience team: founder◦ Windows Security Assurance team: founder◦ Windows Core Security: group program manager◦ Microsoft Passport: group program manager◦ Microsoft Passport User Experience team: manager◦ MSN-client: security and privacy team founder

AT&T Bell Laboratories, IBM, H.E.L.P., LexisNexisPatents 11 US patents, 6 international patents,104 patent citationsEducation Dartmouth College, Rutgers College

Speaker Biography

Page 3: Usable security- It isn't secure if people can't use it. O-ISC conference 14mar2012

Insight Research

Innovation Design

Impact Evaluation

M&A UXMerger & Acquisition User Experience

PI UXProduct Integration User Experience

Strategic UXStrategic UX Management

Sec UXSecurityUser Experience

Product UXDesign & Improve Product UX

User Experience(UX)

Page 4: Usable security- It isn't secure if people can't use it. O-ISC conference 14mar2012

Problem: If a security system wasn’t designed to be usable by each person who touches it, then the people create vulnerabilities

Solution: An end-to-end UX approach that merges technology possibilities, business imperatives, and a deep knowledge of users to improve security

Next Steps: Practical steps to a UX approach

Agenda

Page 5: Usable security- It isn't secure if people can't use it. O-ISC conference 14mar2012

Problem

Page 6: Usable security- It isn't secure if people can't use it. O-ISC conference 14mar2012

The problems with people

Limited decision making skill

Limited number crunching

Emotional responses

Limited Memor

y

Limited ability to visualize

Easily deceived

Limits to vigilance

“Imperfect” cognitive models

Cognitive biases

Too busy

Not tech

savvy

Don’t understand

security

Fear negative outcomes

Don’t respond quickly enough

Lazy

Page 7: Usable security- It isn't secure if people can't use it. O-ISC conference 14mar2012

“The system would be secure if we just got rid of the people.”

Every IT person who ever worked on security

Conclusion

Page 8: Usable security- It isn't secure if people can't use it. O-ISC conference 14mar2012

That is not an option

It is a lot easier to change the system than to change people

Rebuttal

Page 9: Usable security- It isn't secure if people can't use it. O-ISC conference 14mar2012

If a system is not designed to be usable by the people who have to use it, the people are not to blame

The system is

Who is to blame?

Page 10: Usable security- It isn't secure if people can't use it. O-ISC conference 14mar2012

Dialog boxes and vigilance If an end-user sees a security dialog 100 times,

they agree without reading the 101st time Passwords and memory

If a person has to have a 15 character password that must change every 30 days and must contain special characters, they write the password on a Post-it note

Trojans and decision making If a user opens an Excel spreadsheet without

questioning the source, they invite hidden exploits

Examples of People’s Limitations

Page 11: Usable security- It isn't secure if people can't use it. O-ISC conference 14mar2012

It is not just end-users but every human in the end-to-end system

End-to-end

Page 12: Usable security- It isn't secure if people can't use it. O-ISC conference 14mar2012

End-users Product Managers Business Analysts System Designers Program Managers Project Managers Developers Testers Marketing Sales

End-to-end

Installers Administrators Hackers Trainers Maintenance Monitoring Forensics Deprecation etc.

Page 13: Usable security- It isn't secure if people can't use it. O-ISC conference 14mar2012

Developer If a developer does not have insight into the

security skills of the user, they assume the user is like them

Installer If it is too hard for an installer to figure out how to

configure security, it goes in with a risky default Sales

If a sales person can’t model a customer’s security needs sufficiently, they sell them the wrong system

Examples of Non-end-user Limitations

Page 14: Usable security- It isn't secure if people can't use it. O-ISC conference 14mar2012

Am I exaggerating?Severity Test

Page 15: Usable security- It isn't secure if people can't use it. O-ISC conference 14mar2012

Comodo Cert Auth◦Problem: tricked into issuing fraudulent certs

◦UX: people are easily deceived◦Result: employees were socially engineered

Recent Examples

Page 16: Usable security- It isn't secure if people can't use it. O-ISC conference 14mar2012

DigiNotar ◦Problem: hacker access to cert issuing◦UX: people can’t perceive patterns over broad data

◦Result: breach not in admin awareness◦UX: people susceptible to impact bias; a cognitive bias of estimation

◦Result: did not prepare a user scenario for cert revocation

Recent Examples

Page 17: Usable security- It isn't secure if people can't use it. O-ISC conference 14mar2012

Sony◦Problem: data breach 77 Million ID thefts◦UX: people susceptible to confirmation bias

◦Result: did not perceive risk and made poor security choices, insufficient maintenance of patches

◦UX: overconfidence in decision making◦Result: provoked the hacker community

Recent Examples

Page 18: Usable security- It isn't secure if people can't use it. O-ISC conference 14mar2012

RSA◦Problem: token information hacked◦UX: limited ability to predict consequences

◦Result: people post info in social media

◦UX: people are easily deceived◦Result: fooled by phishing attack with Adobe-Excel exploit

Recent Examples

Page 19: Usable security- It isn't secure if people can't use it. O-ISC conference 14mar2012

H.323 Protocol◦Problem: ~150,000 corporate video systems set to auto-answer allowing spying

◦UX: status quo bias◦Result: system default configuration implications overlooked

◦UX: risk assessment skills◦Result: not deployed within secure corporate networks

Recent Examples

Page 20: Usable security- It isn't secure if people can't use it. O-ISC conference 14mar2012

Solution

Page 21: Usable security- It isn't secure if people can't use it. O-ISC conference 14mar2012

Improve end-to-end system security by taking a UX approach to design and development

Take the UX approach

Page 22: Usable security- It isn't secure if people can't use it. O-ISC conference 14mar2012

Many UX TechniquesInsight Research

Innovation Design

Impact Evaluation

Customer InsightUser ResearchIdeationWorkflowTask flowActivity CyclesPain pointsTouch pointsJourney mapEtc.

DesignUser-friendlyInteraction designInformation ArchTransformationSpecificationDesign guidelinesLook and FeelDevelopment Etc.

Usability testingA/B testingCustomer validationBeta testingAnalyticsEvaluationMeasurementsIterationsEtc.

Page 23: Usable security- It isn't secure if people can't use it. O-ISC conference 14mar2012

Insight Research: Detailed attention to the needs, limitations, and behaviors of people in a system to gain insights

Innovation Design: Apply this insight to intentional design in all stages of development, implementation, and use for specific user types

Impact Evaluation: A multi-stage approach requiring analysis, design, and evaluation iterations to ensure successful improvement

UX Approach

Page 24: Usable security- It isn't secure if people can't use it. O-ISC conference 14mar2012

Deeply studying the people in the system Gathering insight into their skills,

motivations, limitations, behaviors, etc. Using that information to drive innovative

designs for security problems

Insight Research

Page 25: Usable security- It isn't secure if people can't use it. O-ISC conference 14mar2012

Keep all users in mind when designing systems

Use the deep insights about users to match design to their limitations and behaviors

Designing to address user pain points and limitations

Innovation Design

Page 26: Usable security- It isn't secure if people can't use it. O-ISC conference 14mar2012

Test with people in the real world not theoretical ideal world conditions

Iterate improvement, evaluate, insight, design cycles◦ UX is an ongoing, incremental approach that

depends on data

Impact Evaluation

Page 27: Usable security- It isn't secure if people can't use it. O-ISC conference 14mar2012

Problem: A security IT tool was not being adopted UX Action: Ethnographic research and contextual

inquiry on the variety of IT people using this security system to determine root cause

Result: Identified 4-5 distinct IT persona types for each of four company IT segments: enterprise, large, medium, and small groups

Separated roles from titles, skills, motivations, and activity/behaviors

Solution: One-size fits all was not working for any group, segmented core product into company/role specific products

Persona Example

Page 28: Usable security- It isn't secure if people can't use it. O-ISC conference 14mar2012

Problem: Significant implementation and customization errors on install and administration

UX Action: Usability study of system with representative users. Included a UX assessment of technical writing.

Result: Root cause was both product interface and the training/documentation

Solution: Improved interaction and improved documentation and training to reduce errors

Usability Testing Example

Page 29: Usable security- It isn't secure if people can't use it. O-ISC conference 14mar2012

Problem: System configuration taking too long and requiring repeated revisions

UX Action: UX evaluation of configuration process

Result: Total over 3,000 configuration options, 6 that system developers could not tell apart, detachment between desired outcome and configurations

Solution: Reduced configuration complexity, options based on real use, aligned outcomes with options, created profiles, offered service

UX Evaluation Example

Page 30: Usable security- It isn't secure if people can't use it. O-ISC conference 14mar2012

Problem: Client with ~900,000 users globally; vendors, employees, on variety of devices, no easy way to see network security status

UX Action: Reviewed current system, modeled pattern of monitoring workflow, prioritized events into semantic map for this audience

Result: Needed situational awareness drill down from simple to detailed, not event alerts

Solution: Created visualizations for quick overall system status with 4 layers of drill down to improve awareness

Visualization Design Example

Page 31: Usable security- It isn't secure if people can't use it. O-ISC conference 14mar2012

Problem: Users relying on password customer support on failed logins◦ Wanted to minimize user frustration◦ Wanted to separate real users from non-users◦ Wanted to minimize customer support costs

UX Action: Researched a variety of real user behaviors to determine optimum design to meet goals

Behavioral Analytics Example

Page 32: Usable security- It isn't secure if people can't use it. O-ISC conference 14mar2012

PSWD attempt

PSWD attempt

PSWD attempt

PSWD attempt

PSWD attempt

PSWD attempt

PSWD attempt

PSWD attempt

PSWD attempt

PSWD attempt

PSWD attempt

PSWD attempt

PSWD attempt

PSWD attempt

PSWD attempt

PSWD attempt

PSWD attempt

PSWD attempt

PSWD attempt

PSWD attempt

PSWD attempt

PSWD attempt

PSWD attempt

PSWD attempt

PSWD attempt

PSWD attempt

PSWD attempt

PSWD attempt

PSWD attempt

PSWD attempt

PSWD attempt

PSWD attempt

PSWD attempt

PSWD attempt

PSWD attempt

PSWD attempt

PSWD attempt

PSWD attempt

PSWD attempt

PSWD attempt

PSWD attempt

PSWD attempt

PSWD attempt

PSWD attempt

PSWD attempt

PSWD attempt

PSWD attempt

PSWD attempt

PSWD attempt

PSWD attempt

PSWD attempt

PSWD attempt

PSWD attempt

PSWD attempt

PSWD attempt

PSWD attempt

PSWD attempt

PSWD attempt

PSWD attempt

PSWD attempt

PSWD attempt

PSWD attempt

PSWD attempt

PSWD attempt

PSWD attempt

PSWD attempt

PSWD attempt

PSWD attempt

PSWD attempt

PSWD attempt

PSWD attempt

PSWD attempt

PSWD attempt

PSWD attempt

PSWD attempt

PSWD attempt

PSWD attempt

PSWD attempt

PSWD attempt

PSWD attempt

PSWD attempt

PSWD attempt

PSWD attempt

PSWD attempt

Average Success Trial of Forgetters with no Lock Out, No CS, and no Self Help

Average Abandon Trial of Forgetters with no Lock Out, No CS, and no Self Help

Average Call if have CS Link

Average Self Help if have Self Help Link

Average CS Call if have CS Link and Self Help Link

Average Self Help if have Self Help Link and Lock Out @ 3

Purple add Self Help Link, Blue add CS link, Yellow you know you’ve got a hacker

Page 33: Usable security- It isn't secure if people can't use it. O-ISC conference 14mar2012

Problem: Client with some divisions having repeated auth setup issues while others didn’t

UX Action: Compared SOP, reports of use, with actual use patterns

Result: Some divisions had activity cycles of use and complete non-use based on business cycle. Start of each cycle users forgot and created issues

Solution: Redesign system for infrequent use to make more intuitive, require users to have refresher when return

Workflow Mapping Examples

Page 34: Usable security- It isn't secure if people can't use it. O-ISC conference 14mar2012

UX approach is not a substitute for good security technology engineering, it is an addition

You have to do both Keep advancing security technologies

Not exclusive

Page 35: Usable security- It isn't secure if people can't use it. O-ISC conference 14mar2012

Next Steps

Page 36: Usable security- It isn't secure if people can't use it. O-ISC conference 14mar2012

Add a UX approach to your security improvement plans

If you have a specific UX-based security problem◦ Develop a tailored UX initiative

If you DO NOT have a specific UX-based security problem◦ Introduce a UX approach in steps

Next Steps

Page 37: Usable security- It isn't secure if people can't use it. O-ISC conference 14mar2012

Start your UX approach today1. Implement: Start with the UX basics2. Design: Adopt and tailor known UX solutions to

fit your situation3. Evaluation: Specifically evaluate your UX

problems, your users, your environment of use, etc. and implement specific solutions

4. Research: Invest in long-term research into the people in your system to drive deep UX understanding

Incremental Steps

Page 38: Usable security- It isn't secure if people can't use it. O-ISC conference 14mar2012

If we all take a UX approach to security system design and improvement, their real-world security value will increase

Take the UX approach

Page 39: Usable security- It isn't secure if people can't use it. O-ISC conference 14mar2012

Darren Kall [email protected] http://www.linkedin.com/in/darrenkall @darrenkall +1 (937) 648-4966

SecUX: We’re glad to help your company have more usable security.

Thank You