For our research it is mandatory to do a research about other researches, so we read a lot of papers which could be important for our topic and literature about latency. Main questions are: How did other researchers conduct their studies? What did they find out? Which results are important for us?

Because latency is a very wide field of research, we classified the topics.

Before investigating about latency, it is important to know what exactly latency is and how it gets defined. The shortest definition is probably: latency is delay. With the term ‘latency’ we usually mean the so called end-to-end latency, which is […] commonly defined as the duration between a user action (e.g. movement of an input device or human limb) and the corresponding on-screen visual feedback (e.g. cursor movement, object displacement) (Casiez et al., 2015, page 629). ‘ ‘Depending on its amount and variation over time (jitter), the end-to-end latency can degrade the action-perception loop, impact user performance and make the system feel less responsive and interactive (Casiez et al., 2017, page 29)’. As Andrea stated in her literature review, ‘[…] latency seems to be an easy to grasp topic [at first glance], whereas upon further investigation it turns out to be complex: Several different aspects in the signal […] influence overall latency (Fischer, 2018) .’

The first regarded aspect of latency is the input lag. Wilson (2009) says that this ‘[…] is an unavoidable reality that can only be minimized and never eliminated (Wilson, 2009).’ He defines it as the time between the user action and the corresponding event arriving at an application (Wilson, 2009).

Another aspect of latency is the network latency or, how Martens et al. define the processing time of the computer running an application, ‘system latency’ (Martens et al., 2018). Other aspects of latency play an important role in online games, for example the latency of the other players and the bandwidth of the internet connection.

Now we know what latency is, we need to understand how it gets measured.

Bockes et al. provide a Raspberry-Pi-based tool before measuring the latency of USB-connected input devices: ‘LagBox’. This system can measure latency with sub-millisecond accuracy and can help identify the sources of input lag in order to reduce end-to-end latency. Bockes et al. not only write about latency itself, but as well about latency variance. They tested several input devices (gamepads, keyboards, mice) and collected 5000 samples for each device.

Casiez et al. provide two methods for measuring end-to-end latency. The first one is ‘[…] a low cost method to measure and characterize the end-to-end-latency when using a touch system […] or an input device equipped with a physical button (Casiez et al., 2017, page 29).’ They describe the approach of measuring the end-to-end latency by using an external camera, in the best case a high frame rate one, for recording the physical action of the user and the on-screen response. After that, the experimenter has to go through the video frame by frame and ‘[…] then count the number of frames between the two events to get one measure latency (Casiez et al., 2017, page 30).’ This method is time consuming but the most commonly used. The second one is a simple method for measuring end-to-end latency using an optical mouse. A standard computer mouse gets positioned at a fixed on a horizontally oriented monitor where ‘[…] a given texture is displayed on the screen and moved of a controlled distance […] while a timestamp is recorded (Casiez et al., 2015, page 631).’ This movement gets recognized as a normal event by the mouse sensor and the timestamps get recorded. In this way, latency can be measured (Casiez et al, 2015).

For our research it is interesting, how much latency is too much latency. Vitus and another student of the class - Ariane Demleitner - wrote their literature reviews about latency and Fitt´s Law. Fitt´s Law is a model of human movement and '[…] predicts that the time to acquire a target is logarithmically related to the distance over the target size(MacKenzie, 1991, page 161)'. As we described in a previous blog entry, Scott MacKenzie presents in his paper “A comparison of input devices in elemental pointing and dragging tasks” an experiment where each task is modelled after Fitt´s Law and the three devices mouse, tablet and trackball get compared (MacKenzie, 1991).Demleitner summarizes in her paper the study design of MacKenzie in a 2D mouse-pointing task: ‘Within this study, the experimental design was conducted with changes in visual latency levels (8.3ms, 25ms, 75ms, 225ms), distances between targets (96px, 192px, 384px), targets widths (6px, 12px, 24px, 48px) and task complexities. The results indicate that delays clearly reduce the user performance when testing motor-sensory movement activities - measurements show that the decrease of human performance starts at 75 ms latency (Demleitner, 2018, page 2).’

The choice of an adequate questionnaire was described in the blog entry about planning the main study. We chose the Self-Assessment Mankin (SAM) which works like a likert scale but non-verbal pictorial. It measures three dimensions: pleasure, arousal and dominance (Bradley and Land, 1994, page 49). 'SAM ranges from a smiling, happy figure to a frowning, unhappy figure when representing the pleasure dimension, and ranges from an excited, wide-eyed figure to a relaxed, sleepy figure for the arousal dimension. The dominance dimension represents changes in control with changes in the size of SAM: a large figure indicates maximum control in the situation (Bradley and Land, 1994, page 50f).'

A field on which latency has an important impact, are games. This was also the topic of Andrea’s literature review where she found out, that ‘Mark and Kajal Claypool have provided a solid foundation of research regarding latency in games, being cited by the majority of other researchers in this field (Fischer, 2018).’ They classify games in the following categories: first person avatar, third person avatar and omnipresent (Claypool & Claypool, 2006) and provide an overview about the different phases of a game and the impact of latency on each is given (this is listed very shortly without further explanations). During the first two phases - setup and synchronization - the players do not get affected significantly by latency. The most important part of a game is the play phase, where latency impacts the player actions and the gaming experience. The transition phase is not affected by latency as well. For a better understanding of their study, which is summarized in the following, there is a short explanation what the two player actions are. The first one - deadline - is the time an action takes to complete and the second one - precision - is the accuracy needed by the player for that action (Claypool & Claypool, 2015).

A bachelor thesis of Mark Claypool’s students (Christopher Burgess, Nathan Roy Quantifying the Effect of Latency on Game Actions in BZFlag) is about measuring the effects of latency on gameplay in an online game. For this, the Game BZFlag was manipulated. This experiment is described in the paper ‘Latency can kill: precision and deadline in online games’ (Claypool & Claypool, 2010). ‘[Their] approach to evaluate and empirically validate [their] precision & deadline model […] and insights […] was to modify an open source, online game to allow for controlled precision & deadline experiments over a range of latencies (Claypool & Claypool, 2010, page 5) ’.

Besides this game, another game was manipulated: Saucer Hunt (Claypool et al., 2015). ‘The player was AI controlled, the game set to “headless” and the frame time set to 0 in order to run the experiments in the background. All combinations of the weapon for speeds 0.25, 0.5, 0.75, 1, 2, 3 and 4, and areas of effect 0, to 10 were tested, a total of 77 combinations. For each combination, delays from 0 to 990 milliseconds were tested in steps of 33 milliseconds (one game loop). For each weapon configuration at each latency, one-thousand games were played for each combination. In all, about sixty-thousand hours of gameplay were emulated, or nearly seven years straight of playing Saucer Hunt (Casiez et al., 2015, page 4).’

Claypool & Claypool, 2016, page 12

In their paper ‘On Latency and Player Actions in Online Games‘ they present the following graphic and sum their results up: ‘[…] Above the grey region, quality is generally acceptable while below the gray region, quality is generally unnacceptable [sic!] (Claypool & Claypool, 2016, page 12).’ In their paper ‘Latency can kill: precision and deadline in online games’ they stated acceptable limits to response times of the system. They claim that ’[a] common conception among game players is that network latencies below 100 milliseconds are essential for unimpaired game play, with maximum tolerable latencies being just over 100 milliseconds, regardless of the game genre (Claypool & Claypool, 2010, page 6).’

• Bockes et al. (2018): LagBox – Measuring the Latency of USB-Connected Input Devices.
• Bradley and Lang (1994): Measuring emotion The self-assessment manikin and the semantic differential.
• Burgess & Roy (2009): Quantifying the Effect of Latency on Game Actions in BZFlag
• Casiez et al. (2015): Looking through the Eye of the Mouse: A Simple Method for Measuring End-to-end Latency using an Optical Mouse.
• Casiez et al. (2017): Characterizing Latency in Touch and Button-Equipped Interactive Systems.
• Claypool & Claypool (2006): On Latency and Player Actions in Online Games.
• Claypool & Claypool (2010): Latency can kill: precision and deadline in online games.
• Claypool & Claypool (2015): A taxonomy for player actions with latency in network games.
• Demleitner (2018): A Review of Fitts' Law and Latency of Input Devices.
• Fischer (2018): Latency in Games.
• MacKenzie (1991): A Comparison of Input Devices in Element Pointing and Dragging Tasks:
• Maierhöfer (2018): Latency and Fitts' Law.
• Martens et al. (2018): Effects of low-range latency on performance and perception in a virtual, unstable second-order control task.
• Wilson (2009): Exploring Input Lag Inside and Out.