Chapter 10

Quality of Service — Team Maui

10.1 Introduction

  • Perception of Computer Speed
    • During the 1960s, computer speed was determined by the response time for mathematical computations, programming compilations, or database searches.
    • Today, because of the emergence of the World Wide Web, users have to take into account network connection speed, server load, and network congestion.
    • They also need to understand the source of errors
      • Before the mass use of PC networking, the source of the errors was always the computer or directly connected devices.
      • Today users have to take into account the multiple sources that could have caused the errors to occur, such as dropped connections, unavailable web servers, and network outages.
    • This complex set of concerns is discussed under the general term Quality of Service (QoS)
  • Concerns over QoS
    • Stem from a basic human value: Time is precious
      • Delays that interfere with task completion such as, lengthy or unexpected system responses can lead to:
        • Frustration, Annoyance and eventually Anger
          • These user reactions then lead to frequent errors and low satisfaction
    • QoS also takes into account a second human value: Harmful mistakes should be avoided
      • There must be a balance between speed and error rates
        • When users work quickly they tend to make errors more frequently
          • This may cause stress for the users, especially if the errors cause data loss, equipment damage, or imperil human life.
        • The pace of work must be slowed to force users to read and comprehend the directions and actions that are present in the system.
    • A third aspect of QoS is to reduce user frustration
      • User frustration usually occurs from:
        • Long, unexpected system delays
        • Crashed that destroy data
        • Software Bugs that produce incorrect results
        • Poor designs that lead to confusion
        • Network issues
  • Network designer, builder, and operator choices dramatically influence the user experience.
    • Web page optimization
      • Byte count and number of files
      • File Previews
      • Choice of fast or slow services (high-resolution, low-resolution)

10.2 Models of Response-Time Impacts

For users, the main experience of Quality of Service is the computer's response time

Response time is the number of seconds it takes from the time a user initiates an action until the computer presents the results.

User think time The number of seconds the user thinks before entering the next action.
Simple Stages of Action Model

  1. initiation of an action
  2. Wait for computer's response
  3. Observe while results appear
  4. Think about results

This simple model is not very realistic because users plan while they are initiating an action (typing/clicking), waiting for results to appear, and interpreting results. Because users are able to use this time to plan, it is very difficult to obtain precise measurements of user think time.

  • Response Time is usually easier to estimate
    • Many times pop-up messages are displayed immediately after the initiation of an action so the response time doesn't seem as long.
    • Delays greater than 160 milliseconds are noticed and become annoying. However, users have come to accept delays from networked devices.

Designers and network managers who seek to provide high QoS have to consider several factors including: technical feasibility, costs, tasks complexity, user expectations, speed of task performance, error rates, and error-handling procedures.

  • Most users prefer rapid interactions
    • Response times that are longer than 15 seconds can be detrimental to productivity, can increase error rates and decrease satisfaction.
    • Rapid interactions, faster than 1 second, can increase productivity, but may also increase error rates for complex tasks.

Web-site display performance was studied by evaluating delay, familiarity, and breadth to examine interaction effects on user performance, attitudes, stress, and behavioral intentions. Lab-tests were conducted to determine "acceptable" delays in two cultures, U.S. and Mexico. Although not all results have been published, conclusions show that user impatience is high, especially in the U.S. as compared to Mexico.

  • Refresh Rates - can lead to user frustration if too slow and can be very pleasing when operating speedily
    • In web applications screen refresh rate may be limited by network transmission or server performance. Images may appear in fragments over several seconds.
  • Network Connection Speeds
    • There are several network options for consumers to purchase from 56-Kbps dial-up to 50-Mbps Fiber Optic Service (FiOS) that will affect the quality of service that the user receives.
    • Internet service providers (ISP) have typically offered plans with a much greater download speed than upload speed because most users download more content than they upload. However, with the current "user-generated" content era, it is increasingly important for upload speeds to keep up with download speeds.
    • There are tools available that allow users the check their connection speeds and have a better idea of the quality of service they are receiving from their current ISP. However, they need to understand that network traffic and server loads can also have an effect on their connection speeds, especially during peak usage times.
  • Reading textual information from a screen is often a challenging cognitive and perceptual task
    • Users tend to relax, pace themselves, and work productively when the screen fills with text instantly
    • Users often scan the documents to find the information they are looking for
    • Because of these facts, it is useful to display text first and leave space for graphics that are slower to load
    • As display quality improves, as more people are going 'green', and as online books and newspapers become increasing available, there is an increasing demand for rapid display of textual and graphical data.
  • Limitations of short-term and working memory
    • Magic number seven - plus or minus two
      • The average person can rapidly recognize seven 'chunks' of information at any given time
        • They can hold this information for 15 to 30 seconds in short-term memory
        • The size of the chunks of information depends on the person's knowledge and experience about the material
      • Performing a distracting task during this time, erases the chunks
      • If a person focuses on retaining the information it can be transferred to long-term memory
    • Short-term memory and working memory are used together to process information and solve problems
      • Short-term memory processes perceptual input
      • Working memory generates and implements solutions
    • People tend to combine several lower-level concepts into a single higher-level chunk to help them remember complex problems
    • Short term and working memory are both highly volatile
      • Disruptions can cause loss of memory
      • Delays may require that the memory be refreshed
  • What causes errors?
    • After a user is able to construct a solution to a problem, he/she must then record or implement the solution.
      • The potential for errors increases and the pace of work slows when the solutions have to be recorded
    • When using an interactive computer system, users formulate plans and then have to wait while each step of the plans are executed. If the execution takes too long or if an unexpected result is obtained, the users may forget a portion of the plan and therefore be forced to continually review the plan. This can cause slowed productivity and more errors.
    • Longer response times cause users to become anxious because the penalty for an error increases
      • When users are anxious, their performance slows and errors increase.
    • Response times that are too short could cause the user to skip or fail to understand important materials or even obtain incorrect results
  • Paced vs. Un-paced tasks
  • Car driving analogy
    • Higher speeds increase the potential for accidents, so speed limits are provided to lower the risk of the dreadful consequences. "When incorrect use of computer systems can lead to damage to life, property, or data, should not speed limits be provided?"
    • Talking on a cell phone while driving has shown to increase accident rates. Computer users who tend to multitask make more mistakes. GPS systems are available to aid drivers in getting to their destination. Agents and wizards guide novice users to successful conclusions, but will this grow in the near future?
  • Progress indicators tend to shorten perceived elapsed time and heighten satisfaction by reassuring the user that the process is underway:
    • Graphical indicators (usually better than static, blinking, or numeric)
    • Blinking messages
    • Numeric seconds left for completion
  • Conditions for Optimum Problem Solving
    • Users can achieve rapid task performance, low error rates, and high satisfaction if all the following criteria are met:
      • Users must have sufficient knowledge of the objects and actions necessary to complete the task
      • The solution plan can be carried out without delays
      • Distractions are eliminated
      • User anxiety is low
      • Feedback is given about progress toward solution
      • Errors can be avoided or easily handled
    • Other conjectures that play a role in choosing the optimum interaction speed
      • Novices prefer to work at slower speeds so they normally exhibit better performance with slower response times
      • When there is little penalty for an error, users prefer to work faster
      • When the task is familiar and easily comprehended, users prefer more rapid action
      • If users have experienced rapid performance previously, they will expect and demand it in future situations

In order to better evaluate user productivity, researchers have extended models to include tempting distractions and unavoidable interruptions, such as arriving e-mail, instant messages, phone calls, and other requests.

10.3 Expectations and Attitudes

  • Response Times
    • What are acceptable response times?
    • The 2 second limit is appropriate for many circumstances, but sometimes a tenth of second is necessary.
      • Ex. 2s -v- .1s response times of dial tones and key presses
  • Factors that influence acceptable response times
    • Expectations
    • User tolerance for delays
    • Task complexity
  • Expectations
    • Once you go broadband, you never go back.
    • If tasks are completed quicker than expected, people will be pleased, and vice versa. (traffic lights)
    • One way to reduce network complaints is by using a response-time choke.
      • When the load is light the system can perform slower to give uniform speed to all users.
      • Disruptions frustrate users who develop a working style based on a certain level of responsiveness.
      • Some users refuse to work when response times are slow.
    • Users expect a rapid startup for laptops/cameras, and are annoyed with waiting times.
  • User tolerance for delays
    • Novices have more patience.
    • Variations in acceptable waiting times.
      • Personality (Laid back/Demanding)
      • Cost (Twitter/Air traffic control)
      • Age (Young/Old)
      • Mood (Carefree/Upset)
      • Cultural context (Mexico/US)
      • Time if delay (1:00pm/4:30pm)
      • Noise (Quiet/Loud)
      • Perceived pressure to complete a task (No pressure/Deadlines)
  • Task complexity
    • Repetiitive tasks demand faster response times.
    • Complex tasks with longer response times allow for users to plan ahead.
    • With complex tasks, users will adapt their working sytle to multitask duing delays.
    • Even so, excessively long delay will cause user dissatisfaction.
  • Tasks that demand rapid system performance:
  1. Video games
  2. Flight simulators
  3. Graphic design
  4. Dynamic queries
  5. VoIP
  6. Streaming multimedia

These tasks require no perceived delay (<100 milliseconds)

Long response times in the WWW lead users to view the company negatively and find the content less interesting. Use of Ajax and other dynamic techniques increase responsiveness and user expectations.

User controlled pace settings can benefit expert users as well as those who are older or disabled. It also allows companies to charge a premium for faster internet service.

  • Summary
  1. User are adaptive and will work faster as they gain experience. Allowing the user to set their own pace can be useful.
  2. Repetitive tasks require a short response time.
  3. With complex tasks users can adapt to slow response times with productivity loss, but dissatisfaction sets in with longer response times.

10.4 User Productivity

Productivity can in many ways be related to the amount of work accomplished in a given period of time. Therefore, in order to improve user productivity, the efficient use of response times is vital.

Shorter response times can lead to increased productivity. However, long response times may give users opportunities to work on concurrent processes, reducing the effort and/or time required to finish a task.
The opposite of this may also occur. Working too quickly because of fast response times, can result in errors, reducing productivity.

Because of these two extremes designers must carefully assess each situation to determine appropriate response times for optimal levels of user productivity. For occasional tasks the importance of this is less critical. But for tasks that occur frequently, determining proper response times should be given greater priority in order to prevent loss of productivity.

An alternative to forcing users to sit through longer response times is to hide the delay. Designers can display important or critical information while other information or processes are still loading. This technique is very useful for websites in order to keep users engaged while they wait for the rest of the content to be loaded in the browser.

The nature of a task strongly influences decisions about adjusting response time. Faster response times can allow users to work more quickly, but decisions may be less than optimal. However, this faster pace might also allow a user to quickly reverse actions and try new ones, reducing the penalty for errors. Users may also learn how to use a system faster since shorter response times would allow for easy exploration of alternatives.

Users adopt various strategies for data entry depending on the response time of a system.

Response Time User Strategies Result
Less Than 1 Second Work without checking to see if the system is ready for input. Increased Errors
Between 1 and 2 Seconds Work is paced. Wait until system is ready for input. Appropriate time is given for the system to accept inputs
Greater Than 2 Seconds Increased monitoring of system Users make sure system is ready for input

For complicated problem-solving, users adapt their work style to the response time. Changes in response time between 0.1 and 5 seconds do not impact productivity because of this. Simple and habitual tasks receive the greatest benefits in productivity due to faster response times.

In the case of complex tasks, shorter response times result in hurried decisions and increased errors. Longer response times increase the frustration a user may already be experiencing while trying to solve a problem, placing a burden on short-term memory.

All of this information points to several things:

  • Users will adjust to the pace of an interface
  • As they adjust, users prefer a faster pace
  • Errors for complex tasks increase with faster response times
  • All tasks seem to have optimal response times that reduce errors
  • The greater the consequence for error, the more important it is to slow users down so they will think

10.5 Variability in Response Time

Consistent and predictable experiences are valued for their ability to reduce anxiety and uncertainty.

If the actual response time differs from a user’s expected response time, the user may become uneasy. Unsettling reactions like this can be prevented (or acknowledged) through the use of indicators and progress reports.

Even moderate variations in response times can be detected by a majority of users. Because providing a standard response time for all individual actions would be difficult, it is suggested that designers specify a general response time for classes of similar or related actions.

These slight variations in response time do not seem to adversely affect performance, as users are capable of adjusting to them in most contexts. It should be noted that this is not a universal observation and a number of users will still be frustrated by variation in response time for some tasks.

For high stress work environments, adjustments to response times can lower the amount of pressure being placed on users. For most environments however, there is no evident link between response time variation and effects on user wellbeing.

To tie it all together:

  • Minor variation in response time is tolerable for most users and has little effect on performance
  • Frustration can emerge if delays are roughly at least twice the expected response time
  • When slow responses cannot be avoided provide information about progress

10.6 Frustrating Experiences

Another thing to think about quality of service is the quality of user experiences. Recent research has begun to document and help us understand the sources of user frustration with contemporary user interfaces.

When computer cause users to become frustrated it can affect:

  • Workplace productivity
  • Users' moods
  • Interactions with other co-workers

The largest number of frustrating experiences occurred while using:

  • Word processors
  • E-mail
  • Web browsers

Frustration levels were strongly related with the amount of time lost/ time required to fix the problem and with the importance of the task.
The majority of users remembered frustrating incidents such as:

  • Incorrect auto formatting
  • Computer errors or bugs
  • Slow or dropped Internet connections
  • Unwanted pop- ups*

These incidents all seem to have a couple of things in common:

  • They are external to the user’s cognitive processing
  • They interrupt the user’s task and take control away from the user

Infrastructure improvements to server capacity and network speed and reliability will improve user experiences, but the continuing growth of Internet usage means there will be problems for many years to come.

E- mail has become the source of frustrating “ spam”. Some of these messages come from major corporations who make an effort to focus their e-mail on current customers, but much spam comes from small companies and individuals who take advantage of the low cost of e-mail to send blanket notices to huge lists of unfiltered e-mail addresses. Anti-spam legislation is being passed in many nations, but the Internet’s international reach and open policies limit the success of legal controls.
Some spam senders and advertisers claim freedom of speech in their right to send spam or ads, but most users wish to see some limitation on the right to send bulk e-mails or unsolicited pop-up ads.

Another frustrating problem for users is the prevalence of malicious viruses which cause various problems and frustration for users. In 2000, before anti-virus software became effective, the famed ILOVEYOU virus contaminated millions of personal computers worldwide by tricking users to open e-mail messages by placing the words “ I Love You” in the subject line; recovering from the damage cost an estimated $10.2 billion.

Since frustration, distractions, and interruptions can impede smooth progress, design strategies should enable users to maintain concentration. Three initial strategies can reduce user frustration:

  • Reduce short-term and working memory load
  • Provide information-abundant interfaces
  • Increase automaticity.

Automaticity in this context is the processing of information (in response to stimuli) in a way that is automatic and involuntary, occurring without conscious control. An example is when a user performs a complex sequence of actions with only a light cognitive load, like a driver following a familiar route to work with little apparent effort.

Unless otherwise stated, the content of this page is licensed under Creative Commons Attribution-ShareAlike 3.0 License