Knock Brush! Perceived Impact of Push-based Notiﬁcations

on Software Developers at Home and at the Ofﬁce

Vanessa Vella

and Chris Porter

Department of Computer Information Systems, Faculty of ICT, University of Malta, Malta

Keywords:

Digital Interruptions, Notiﬁcations, Software Developers, Human Factors.

Abstract:

Responding to a digital interruption requires software developers to transfer their attention from their ongoing

task to the contents of the notiﬁcation: a shift which will disrupt their original ﬂow of work. This paper

distinguishes between different types of notiﬁcations (intrusions and interruptions) and reﬂects upon the results

of a survey with 88 respondents. This study contributes to research by providing insights consistent with

literature, particularly related to how users react to push-based notiﬁcations during varying contexts, how the

software developers perceive the cost and beneﬁts of a notiﬁcation, what features stand out in a notiﬁcation

and the strategies used to continue their work following a notiﬁcation.

1 INTRODUCTION

Push-based notiﬁcations can be distinguished by their

content: how relevant is the message right now; tim-

ing: how convenient is it that this message is deliv-

ered at 10:00; and trigger, referring to the impact this

notiﬁcation will have on the user, the options the user

can choose from, to interact with this notiﬁcation, and

the likelihood that a user will ignore the notiﬁcation

(Morrison et al., 2017).

The advent of digital disruptions has the negative

connotation of disrupting the user’s ﬂow state: the

mental state where the software developer is concen-

trating so deeply on the ongoing task that they fail to

notice the environment around them or how long have

they been working on this task (Gray and Rumpe,

2017). Needless to say, software developers remain

fully aware and focused on their primary task at hand

(Okoshi et al., 2017). Nevertheless, the greater issue

of digital disruptions is the ability to ﬁnd a balance

between signal and noise, or in other words having

the ability to granularly deﬁne what an important and

actionable notiﬁcation consists of, while disregard-

ing the less important notiﬁcations. Having achieved

this state of balance will in turn help the software de-

veloper lengthen their attention span thus improving

their ability in performing long-term work (Okoshi

et al., 2017).

https://orcid.org/0000-0003-0654-7237

https://orcid.org/0000-0003-3813-2941

Whilst researchers agree that working in an over-

whelming state of noisy interruptions comes at a

detriment to productivity, literature has also shown

that knowledge workers tie a quicker response time

in handling notiﬁcations to better interpersonal skills

(Fitz et al., 2019). This hypothesises that the ability

to handle notiﬁcations will in turn be seen as having

better task handling by management. For this reason,

knowledge workers are more inclined to be aware of

any notiﬁcations that they may receive to further con-

tribute towards their social accomplishments (Okoshi

et al., 2017).

With regard to the study methodology, a quantita-

tive empirical study was implemented, where a sur-

vey was drawn with the purpose of acquiring data re-

lated to the perceived impact of a desktop digital in-

terruption on a software developer. A mix of open-

ended and likert-scale questions were used with the

aim to best capture the day-to-day life of a knowledge

worker and how the software developer is impacted

by a notiﬁcation, particularly delving into both the di-

rect and mediated effects, if any.

The paper is split up as follows: the next section

is an overview of related literature and the deﬁnitions

adopted in this study. Section 3 details the implemen-

tation of this study, while section 4 contains a thor-

ough discussion of the results of this study. Section 5

discusses the study limitations and possible directions

for future research. A conclusion section (section 6)

presents a synthesis of the main ﬁndings.

214

Vella, V. and Porter, C.

Knock Brush! Perceived Impact of Push-based Notiﬁcations on Software Developers at Home and at the Ofﬁce.

DOI: 10.5220/0010298002140221

In Proceedings of the 16th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications (VISIGRAPP 2021) - Volume 2: HUCAPP, pages

214-221

ISBN: 978-989-758-488-6

2 BACKGROUND AND

LITERATURE REVIEW

2.1 Digital Interruptions

Literature has shown that there are different types of

digital interruptions, which can be classiﬁed as fol-

lows (Addas and Pinsonneault, 2015):

• IT Intrusions: interruptions which are of a per-

ceived, induced or mediated nature; for instance

any external event which is irrelevant to the ongo-

ing task. This is further structured as follows:

– Informational Intrusion: a digital interruption

having content related to a secondary task,

– Actionable Intrusion: a digital interruption

which causes the user to stop working on their

task and start working on a secondary task,

– System Intrusion: a digital interruption which

indicates an error in the system or the lack of

availability of system resources; this will thus

disrupt the user’s ﬂow of work.

• IT Interventions: any interruption which disrupts

a knowledge worker’s ongoing task since it re-

veals a discrepancy between what is expected and

what the user will deliver. IT Interventions can be

further split as follows:

– Informational Intervention: a digital interven-

tion which contains data that is directly relevant

to the user’s primary task,

– Actionable Intervention: a digital interruption

that instigates two-way communication and is

directly related to the ongoing task: most often

actionable interventions reduce the discrepancy

between what the user needs to ﬁnish this task

and how to ﬁnish this task.

Hence, a user can respond to a notiﬁcation in four pos-

sible ways (Mehrotra et al., 2016):

• Handle it immediately,

• Acknowledge it, but keep working on their task,

and handle it later,

• Decline it, that is, explicitly close the notiﬁcation,

• Withdraw it, that is, implicitly refuse to handle it.

2.2 Task Analysis

In Human-Computer Interaction (HCI), task analysis

can be deﬁned as ”creating a sufﬁciently rich and ab-

stract representation of a task situation as it exists or

might exist to allow reasoning about the task situa-

tion without having to refer to the actual task situa-

tion itself.” (Haan, 1998) This deﬁnition directly en-

compasses the fact that interaction in HCI often times

occurs at a high-level of abstraction.

Consequently, multi-task analysis can be deﬁned

in terms of the primary and secondary tasks it is made

up of; a primary task is the task which takes priority,

meaning the task which requires the larger amount of

mental and physical resources (de Haan et al., 1991).

Meanwhile, a secondary task is a smaller task which

is related to the primary task’s scope, and completion

of this task, directly supports completion of the pri-

mary task (de Haan et al., 1991).

2.3 Cognitive Load in Software

Cognitive load is a measure of the amount of ef-

fort being used by the working memory — working

memory is the part of the cognitive system that pro-

cesses the information regarding the user’s primary

task (Kalyuga and Singh, 2016). Working memory

is prone to an overload, meaning that it is suscepti-

ble to discarding information if it is no longer being

processed. Software development tasks such as pro-

gramming and code debugging are dependent on the

working memory. Often times, the software devel-

oper will have to abstract an algorithm and think of

the next steps, whilst already making plans to refactor

the code that they have developed earlier. This pro-

cess is referred to as dividing and conquering by Pe-

tre et al. (2016) who believe that software developers

essentially address smaller problems independently,

whilst still reﬂecting on the associations between dif-

ferent code blocks. Consequently, developers are able

to make adjustments in different code points in scope

of the larger task. Petre et al. (2016) then proceed to

notice that high-performing teams often collaborate in

order to identify, as early as possible, any divergences

in speciﬁcations that might occur if one team makes

changes whilst the other does not.

Maintainable code is code that follows practices,

conforms to established design patterns and is read-

able (Code, 2018). Software developers are accus-

tomed to reading code and building mental models

depending on what they are developing. The men-

tal models are in turn converted to code representa-

tion when the developer manages to implement the

changes. This process is heavily dependent on the

software developer’s abstraction skills, that is, code

reading is a learning process. Therefore, if the code is

understandable, the knowledge worker is said to make

use of less extrinsic load, and consequently lessen

cognitive load (Kalyuga and Singh, 2016). By deﬁ-

Knock Brush! Perceived Impact of Push-based Notiﬁcations on Software Developers at Home and at the Ofﬁce

215

nition, extrinsic load is one of three subparts of cog-

nitive load and refers to the unnecessary work that

the learning process is making use of — which could

have easily been allocated to the learning process it-

self. The other subparts are (a) intrinsic load, which

refers to the complexity of the learning process itself

and the user’s ability in gaining new knowledge, and

(b) germane load, which is the effort required to store

sub-parts of the learning process in long term memory

(Kalyuga and Singh, 2016).

2.4 Current Implementation of

Notiﬁcations in Operating Systems

All three major operating systems (OSs) have a sim-

ilar implementation for push-notiﬁcations, meaning,

allowing installed applications to interrupt the user

with a message when an event occurs, new data is

available, or the status of an object has changed. Nev-

ertheless, there are slight differences in the imple-

mentation and conﬁguration between the OSs. For

instance, notiﬁcations on macOS (Apple, 2020) are

shown at the top-right corner of the screen, and listed

in the notiﬁcation centre. An actionable notiﬁcation

in macOS is called an alert, while an informational

one is called a banner (Apple, 2020). Notiﬁcations in

Windows (2020) are shown at the bottom-right cor-

ner of the screen, and listed in the action center; these

can be either actionable or informational, depending

on the application which pushed it. Notiﬁcations in

Linux, particularly GUI based implementations such

as Ubuntu, also show informational notiﬁcations at

the top-right, but so far have not implemented action-

able notiﬁcations (Ubuntu, 2020). A recent notiﬁca-

tion list is also available in Ubuntu (2020).

One important feature related to notiﬁcations

which is implemented in all three mentioned OSs is

the Do Not Disturb feature – as labelled in macOS

(Apple, 2020) and Linux (Ubuntu, 2020), while la-

belled Focus Assist in Windows (2020).

3 SCOPE AND METHODOLOGY

The general objective of this research is to understand

how desktop notiﬁcations affect knowledge workers

during a state of ﬂow. Consequently, this study fo-

cuses on the daily use of a desktop system from the

GUI (Graphical User Interface). Thomas Davenport

(2005) deﬁnes knowledge workers as individuals who

”have high degrees of expertise, education, or experi-

ence, and the primary purpose of their jobs involves

the creation, distribution, or application of knowl-

edge.” Davenport (2005) continues to argue that by

considering software developers as knowledge work-

ers, implicitly development activities are thereby re-

garded as knowledge creation processes. Based on

the above, and for the scope of this research, we limit

our deﬁnition of knowledge workers to software de-

velopers.

For this reason, this study aimed at building a bet-

ter ﬁrst-hand understanding of the different notiﬁca-

tion taxonomies. Following a critical review of re-

lated literature while also familiarising better with the

domain, a survey was drawn in order to acquire the

perceptions and experiences related to digital inter-

ruptions. The survey questions were designed to en-

courage, as much as possible, software developers to

open up on their experiences of interruptions during

immersive tasks and how they dealt with picking up

from where they left off. For this reason, qualita-

tive questions were used, at times supplemented with

quantitative probes on speciﬁc aspects as informed

from existing literature.

In terms of resources, this survey made use of a

freely available platform: Google Forms. The sur-

vey administered the informed consent and a brief

overview of the scope of the study. No personal iden-

tiﬁable data was collected, however a number of ques-

tions were included related to experience in the ICT

sector. In terms of distribution mechanisms, the sur-

vey was sent as an online link to a number of ICT

companies, as well as shared on social media, includ-

ing Facebook and LinkedIn.

4 FINDINGS

In this section the survey results will be collated, dis-

cussed and compared with the relevant literature.

4.1 General Observations

88 respondents took the time to answer the survey,

two of which contain spam data and will be excluded

from the results, therefore bringing the total number

of participants down to 86.

The participants come from varying ICT back-

grounds with 14 participants (16.3%) having less than

a year of experience in the ﬁeld. Conversely, 21 par-

ticipants (24.4%) have more than 7 years of experi-

ence in the ICT sector. The rest of the participants

were in the 1-3 (25.6%) and 4-6 (33.7%) years-of-

experience brackets.

The majority of the participants (81%) speciﬁed

that they used Windows as one of their development

platforms, if not the only one, followed by Linux

(29.8%) and macOS (14.3%).

HUCAPP 2021 - 5th International Conference on Human Computer Interaction Theory and Applications

216

The most common examples of desktop notiﬁca-

tions received by the software developers were new

email notiﬁcations (86%) and instant message chats

(83.7%) from communication channels. Other ex-

amples of push-notiﬁcations include system updates

(75.6%), task updates from productivity systems such

as Jira (47.7%), and notiﬁcations unrelated to work,

including phone calls, third-party weather sites, arti-

cles, and social media (38.4%). The platform which

is most frequently used to communicate with col-

leagues is emails as a structured platform and Mi-

crosoft Teams for instant-messaging. Table 1 con-

tains a breakdown of the mentioned platforms; note

that the participants had the option to choose multiple

services.

Table 1: Communication platforms mentioned by the par-

ticipants, ordered by popularity descending.

Platform % of Participants

Emails 69.77%

Microsoft Teams 65.12%

Slack 51.16%

Skype 40.70%

WhatsApp 40.70%

Zoom 25.58%

Google Meet 22.09%

Discord 17.44%

Facebook Messenger 17.44%

Google Hangouts 12.79%

Phone Calls 3.49%

Rocket Chat 2.33%

Lifesize 1.16%

LinkedIn 1.16%

Teamviewer 1.16%

Telegram 1.16%

4.2 Do Not Disturb Usage

The participants were asked whether they make use of

the Do Not Disturb feature and more than a third of

the software developers speciﬁed that they do occa-

sionally use this feature (36%). 29 developers (34%)

speciﬁed that they make use of the feature quite reg-

ularly, while 26 (30%) reported that they never make

use of this feature. When probed further to specify

the scenario where the feature is switched on, almost

a quarter of the participants mentioned meetings and

screen-sharing occasions (25.6%), and when the fre-

quency of notiﬁcations suddenly increases due to, for

instance, a spike in group chat activity. Participants

also mentioned a number of scenarios where this fea-

ture is used, including deploying code to production,

troubleshooting and debugging code, as well as code

comprehension and reviewing. All of these tasks can

be considered to be immersive in nature, which can

have signiﬁcant side effects if errors are introduced.

This corroborates observations presented in (Baethge

and Rigotti, 2013), which state that interruptions in-

crease stress and cognitive load since these disrupt

the software developer’s state of ﬂow and thus re-

quire the utilization of additional resources and effort

to achieve the goal of the primary task.

4.3 Notiﬁcation Characteristics

It is also important to note that the frequency of in-

stant messaging notiﬁcations is rather high, with 57

participants (66.3%) stating that they receive more

than one instant messaging (IM) notiﬁcation daily,

while 19 (22.1%) receive at least one IM notiﬁcation

every day. For this reason the participants were asked

to specify in what ways do they determine whether a

notiﬁcation has a higher priority. The following is the

list of features identiﬁed:

• Who the Sender Is: particularly if it’s a team

member, the client or contacts marked as VIP (i.e.

ﬂagged so that all communication from this con-

tact passes through regardless of whether do-not-

disturb is switched on),

• The Application Generating the Notiﬁcation:

participants observed that certain applications,

such as Microsoft Teams, are considered to gen-

erate work-related notiﬁcations, while other plat-

forms, such as Messenger, are considered to treat

the social context,

• Notiﬁcation Target: notiﬁcations which repre-

sent direct messages are perceived as more im-

portant than messages in a group chat, similarly

when a developer is “tagged” or mentioned in the

group chat, as opposed to a general message to all

members,

• The Nature of the Notiﬁcation: whether it is in-

formational or actionable,

• Colour Scheme: red icons imply a higher prior-

ity,

• Directly Related to an Ongoing Task or Dead-

line,

• The Blurb: the information contained in the pop-

up including the title and the topic, especially if it

is a production issue,

• Urgency Level: The business priority or status

attributed to the task update (in the case of a task

update), or a ﬂag in Microsoft Outlook emails.

Similarly, when asked how do they determine if a no-

tiﬁcation is of a lower priority, the software develop-

Knock Brush! Perceived Impact of Push-based Notiﬁcations on Software Developers at Home and at the Ofﬁce

217

Figure 1: Stacked bar chart showing how software developers react to digital interruptions during different types of primary

tasks.

ers noted that automated notiﬁcations, system mes-

sages and spam or promotional messages often go un-

noticed or completely disregarded, with little to no

disruption to their state of ﬂow. The above features

list ties in to how important these traits can be in as-

signing priority (Paul et al., 2015). Paul et al. (2015)

argue that interruptive notiﬁcations can decrease a

user’s attention by offering a distracting focal point.

On the other hand, the authors (Paul et al., 2015) also

specify that having a second focal point is not neces-

sarily detrimental, in fact, notiﬁcations are known to

support task handling and decrease user stress levels

by providing more information on the system’s cur-

rent status.

4.4 Impact on the Primary Task

The context surrounding the digital disruption is one

of the larger determinant factors in the user’s deci-

sion as to the way they will tackle it and when. The

participants were asked to distinguish and mark the

ways in which they generally handle notiﬁcations in

varying contexts. One method which seems to be

frequently adopted is the acknowledge and postpone

protocol, which corroborates ﬁndings in (Mehrotra

et al., 2016). This shows that the majority of the

participants take some extra measures and implement

contingency strategies before moving on to handle the

notiﬁcation. A second interesting result to point out is

that the handle immediately protocol (Mehrotra et al.,

2016) is normally performed when the software de-

veloper is working on less critical tasks, including re-

search and documentation. A graphical representa-

tion of these results can be found in ﬁgure 1. In a

follow-up question, all participants agreed that they

will read through notiﬁcations which are related to

the ongoing task, such that they will be able to take

the necessary information into consideration. Con-

versely, digital interruptions which are unrelated to

the primary task are checked for urgency and then dis-

missed if these are of lesser priority. One software de-

veloper also addressed the fact that when stress levels

are high, a notiﬁcation is able to alleviate unnecessary

stress. This is in-line with ﬁndings in literature (Ad-

das and Pinsonneault, 2015) where notiﬁcations are

known to be an aiding sign in identifying when a soft-

ware developer might want to switch to another task.

Furthermore, Paul et al. (2015) also discussed that

software developers require a ”no notiﬁcation zone”

- meaning that they will occasionally need to work

in a context that is sensitive to interruptions. This

state is otherwise known in this study as being in a

state of ﬂow, since the user is working on an immer-

sive task. The majority of the software developers

noted that during coding, especially, when making

good progress, the users ﬁnd themselves in a state

of ﬂow. Interestingly, one software developer ex-

plained the feeling as follows: “Implementing a set of

requirements I understand clearly, using a program-

ming language I am familiar with.” – where the soft-

ware developer is hinting that s/he is feeling prepared

and unless a disruption occurs, the task will be wholly

completed. Other immersive tasks reported by par-

ticipants include initiating larger projects, designing

data models and methodologies, data migration pro-

cesses, production deployments, debugging and trou-

bleshooting, repetitive tasks (”small and mundane”),

prototyping new technologies, creating UI mock-ups

and user-ﬂow diagrams, creating speciﬁcations for a

new task, code reviewing and refactoring, scripting,

identiﬁcation of optimal architectures, and working

on complex algorithms such as thread synchronisa-

tion.

HUCAPP 2021 - 5th International Conference on Human Computer Interaction Theory and Applications

218

Literature shows that digital interruptions do frag-

ment the execution ﬂow of the primary task by ex-

pecting users to switch their attention to the secondary

tasks (Sonnentag et al., 2018). This consequently has

led to the need that software developers implement

contingency strategies so that they will be able to con-

tinue from where they have left off. State saving of

the primary task helps the knowledge worker resume

the suspended tasks quicker (Sonnentag et al., 2018).

While not all software developers implement contin-

gency strategies (18.6%), since they are able to re-

call what the primary task was, the rest of the partici-

pants recalled implementing certain protocols when a

context-switch is required. The following is the list of

methodologies mentioned, while ﬁgure 2 graphically

depicts the number of frequency of participants per

approach mentioned:

(A) Split their work into smaller projects that can be

tackled as a whole, in anticipation of digital in-

terruptions,

(B) Adding short notes (or voice notes) in Microsoft

OneNote, or in the Task Management tool, or

physically on a paper or sticky note,

descriptive commit messages,

(D) Adding comments and TODO comments in the

code,

(E) Leaving the IDE open and include some high-

lighting to know where to pick up again, or pin

code ﬁles,

(F) Before starting to code, the software developer

writes a pseudo-code version of their implemen-

tation, so they can consult it along the way or

after an interruption,

(G) Use multiple desktops, so that the state of one

desktop is not affected,

(H) Purposefully include an error in the code, for ex-

ample write gibberish code so that the software

developer knows on which line of code they were

working on recently,

(I) Organise browser tabs based on which task takes

priority, and pin the ones which need to be at-

tended to later.

Figure 3 shows that the participants speciﬁed that

related instance messages and related emails are less

distracting as opposed to software updates and unre-

lated emails. Nevertheless, conﬂicting results were

found in literature in this regard; Paul et al. (2015)

stated notiﬁcations related to the software developers’

primary tasks provide little to no beneﬁt and can be

disruptive. Nevertheless, the results obtained in this

Figure 2: Frequency chart showing the methodologies

adopted by the participants when a context-switch is re-

quired.

study corroborate Mark et al.’s (2005) work, where

the researchers focused on fragmented workﬂows and

found out that interruptions outside the users’ work

sphere are more disruptive than those related to soft-

ware developers’ primary task.

Figure 3: Likert plot showing how software developers per-

ceive types of notiﬁcations, with 5 being the most disruptive

and 1 being the least.

4.5 Impact on Users’ Emotional State

This study showed that 66 participants (76.7%) in fact

check that all the necessary communication channels

are open and working before they start their work-

day. 14 software developers (16.3%) noted that this

check is occasionally performed, while only 6 (7%)

denied checking. This result suggests that the soft-

ware developers do in fact anticipate having noisy

days, and ensure that all the relevant notiﬁcations do

pass-through. As the number of push-notiﬁcations in-

creases, literature has shown that software develop-

ers have the increased tendency to anticipate interrup-

Knock Brush! Perceived Impact of Push-based Notiﬁcations on Software Developers at Home and at the Ofﬁce

219

tions, thus working harder during notiﬁcation down-

time to compensate for when the cost of distraction is

too high (Petre et al., 2016).

In the same context, the software developers were

asked to specify whether they experience a fear-of-

missing-out (FOMO), which could arise from a re-

duced notiﬁcation-frequency when compared to a

typical workday. Interestingly, only 17 participants

(19.77%) admitted that they feel fearful in such a

case. Similarly, 19 participants (22.09%) speciﬁed

that they do occasionally have feelings of FOMO.

Nevertheless, the majority (56.8%) denied having

FOMO feelings. The portion of the cases who ad-

mitted to having FOMO, or occasionally, were then

asked to reﬂect on the incident. One software de-

veloper admitted that since the notiﬁcation handling

strategy implemented is not perfect, the Chief Tech-

nical Ofﬁcer (CTO) felt that extra attention was given

to ensure that any high priority support task notiﬁca-

tions were monitored for and handled on-time.

Furthermore, a few participants (4.7%) noted that

FOMO is also dependent on the working environment

– the participants argued that working in an ofﬁce

helps in this regard since you can get the occasional

”tap-on-the-shoulder” from a colleague. Meanwhile,

the participants were asked whether they considered

push-notiﬁcations more crucial when working from

home, since they are one of the few ways to shift the

software developer’s focus on higher prioritised tasks.

The majority of the participants (61.6%) do feel that

digital interruptions are critical when working from

home. 16 software developers (18.6%) expressed that

they do occasionally view notiﬁcations as important

in this context, while 17 (19.8%) dismissed the state-

ment altogether. The question that ensued was to un-

derstand what strategies do the software developers

implement to ensure that they will receive all the rel-

evant notiﬁcations when working from home. The

users speciﬁed that they attempt to ensure having fast

and stable internet and VPN connections. Other de-

velopers noted that they implement the same strat-

egy as when working from the ofﬁce, with dedicating

part of the screen to communication channels, occa-

sionally checking the notiﬁcation tray, and switching

off the do-not-disturb (DND) functionality. One soft-

ware developer also noted that raising the system vol-

ume would help in identifying incoming notiﬁcations,

even if they are missed visually (e.g. produced within

a browser tab). Sound is inherently one of the mul-

timodal cues that are often times included alongside

the pop-up and the badge number on the tray icon,

amongst others (Addas and Pinsonneault, 2015).

When asked to list any differences, if any, with re-

gard to non-work related digital interruptions in the

context of working from home, a few software de-

velopers (7%) pondered on the fact that since social

interaction becomes more limited, the software devel-

oper receiving a non-work related notiﬁcation is more

likely to handle it, than ignoring or dismissing it. This

research is inline to a recent study (Ohme et al., 2020)

that has shown that socially-distanced users are mak-

ing more frequent and longer use of news platforms

(54% increase), messaging platforms (64% increase)

and social media applications (72% increase).

5 LIMITATIONS AND FUTURE

WORK

Notiﬁcations are an inevitable part of a knowledge

worker’s workday. As discussed in this study, this

issue needs to take into consideration a number of

different perspectives, including the software devel-

oper’s working context, whether the notiﬁcation con-

tains important information, and the user’s primary

task, amongst others. Given that this study collected

only survey data, there might be the case that the par-

ticipants did not feel encouraged to provide factual

and unbiased answers (Vaske, 2011). While the ques-

tions included were designed to motivate the users to

share their opinions on the subject-matter, not all par-

ticipants may have felt comfortable with sharing, with

some opting to write short answers only.

A second study limitation may be related to the

limited contextual-speciﬁcity of the questions. Since

the survey questions were designed to be generic

enough to accommodate the various roles within the

software development community, it can be the case

that not all participants related well to the same ques-

tions. The goal of this study was to produce overall in-

sights into the impact of notiﬁcations as perceived by

the software development community, however, fur-

ther studies can be devised to investigate particular

roles and related contexts.

Stated impact may differ from actual behaviour

— more speciﬁcally, while users do understand hav-

ing the option to tweak notiﬁcation settings, neverthe-

less, very little is done to maximise their state of ﬂow.

Therefore further studies can dig deeper into software

developers’ actual behaviour when presented with no-

tiﬁcations in different contexts. Possible research

questions may look into the impact of different notiﬁ-

cation types on performance while carrying out tasks

of varying levels of difﬁculty. Studies could look at

how, for instance, actionable intrusions and informa-

tional interventions affect performance across differ-

ent groups of knowledge workers based on criteria

such as experience.

HUCAPP 2021 - 5th International Conference on Human Computer Interaction Theory and Applications

220

6 CONCLUSION

Digital interruptions at the workplace pose a num-

ber of threats to the quality of the work done, stress

levels and cognitive mental workload. This study

contributes to research by providing insights consis-

tent with the cognitive factors and human ergonomics

community literature, in relation to how software de-

velopers perceive digital interruptions and the traits

of when notiﬁcations occasionally aid the user, while

sometimes they disrupt.

This research has shown that notiﬁcations related

to an ongoing task are perceived as less disruptive,

and in fact acknowledged and handled quicker. The

inverse also holds, whereby, notiﬁcations which are

unrelated to the primary task will be viewed as disrup-

tive, consequently having a larger effect on the soft-

ware developer’s state of ﬂow. For this reason, con-

tingency strategies, such as note-taking and code ver-

sioning are crucial towards helping the software de-

velopers continue their primary task.

In conclusion, notiﬁcations deﬁnitely hold a sig-

niﬁcant role in updating knowledge workers on their

information services.

ACKNOWLEDGEMENTS

This research is partially funded by the Endeavour

Scholarship Scheme (Malta), which is part-ﬁnanced

by the European Union - European Social Fund

(ESF).

REFERENCES

Addas, S. and Pinsonneault, A. (2015). The many faces of

information technology interruptions: a taxonomy and

preliminary investigation of their performance effects.

Information Systems Journal, 25(3):231–273.

Apple (2020). Human interface guidelines - notiﬁcations.

Baethge, A. and Rigotti, T. (2013). Interruptions to work-

ﬂow: Their relationship with irritation and satisfac-

tion with performance, and the mediating roles of

time pressure and mental demands. Work & Stress,

27(1):43–63.

Code, W. M. E. (2018). Abstraction layered architecture:

Writing maintainable embedded code. In Software

Architecture: 12th European Conference on Software

Architecture, ECSA 2018, Madrid, Spain, September

24–28, 2018, Proceedings, volume 11048, page 131.

Springer.

Davenport, T. H. (2005). Thinking for a living: how to

get better performances and results from knowledge

workers. Harvard Business Press.

de Haan, G., van der Veer, G. C., and van Vliet, J. C. (1991).

Formal modelling techniques in human-computer in-

teraction. Acta Psychologica, 78(1-3):27–67.

Fitz, N., Kushlev, K., Jagannathan, R., Lewis, T., Paliwal,

D., and Ariely, D. (2019). Batching smartphone notiﬁ-

cations can improve well-being. Computers in Human

Behavior, 101:84–94.

Gray, J. and Rumpe, B. (2017). The importance of ﬂow in

software development.

Haan, G. d. (1998). Chapter 5: Task Analysis for User In-

terface Design.

Kalyuga, S. and Singh, A.-M. (2016). Rethinking the

boundaries of cognitive load theory in complex learn-

ing. Educational Psychology Review, 28(4):831–852.

Mark, G., Gonzalez, V. M., and Harris, J. (2005). No

task left behind? examining the nature of fragmented

work. In Proceedings of the SIGCHI conference on

Human factors in computing systems, pages 321–330.

Mehrotra, A., Pejovic, V., Vermeulen, J., Hendley, R., and

Musolesi, M. (2016). My phone and me: understand-

ing people’s receptivity to mobile notiﬁcations. In

Proceedings of the 2016 CHI conference on human

factors in computing systems, pages 1021–1032.

Morrison, L. G., Hargood, C., Pejovic, V., Geraghty, A. W.,

Lloyd, S., Goodman, N., Michaelides, D. T., Weston,

A., Musolesi, M., Weal, M. J., et al. (2017). The effect

of timing and frequency of push notiﬁcations on usage

of a smartphone-based stress management interven-

tion: an exploratory trial. PloS one, 12(1):e0169162.

Ohme, J., Vanden Abeele, M. M., Van Gaeveren, K.,

Durnez, W., and De Marez, L. (2020). Staying in-

formed and bridging “social distance”: Smartphone

news use and mobile messaging behaviors of ﬂem-

ish adults during the ﬁrst weeks of the covid-19 pan-

demic. Socius, 6:2378023120950190.

Okoshi, T., Tsubouchi, K., Taji, M., Ichikawa, T., and

Tokuda, H. (2017). Attention and engagement-

awareness in the wild: A large-scale study with adap-

tive notiﬁcations. In 2017 IEEE International Con-

ference on Pervasive Computing and Communications

(PerCom), pages 100–110. IEEE.

Paul, C. L., Komlodi, A., and Lutters, W. (2015). Inter-

ruptive notiﬁcations in support of task management.

International Journal of Human-Computer Studies,

79:20–34.

Petre, M., van der Hoek, A., and Quach, Y. (2016). Software

Design Decoded: 66 Ways Experts Think. MIT Press.

Sonnentag, S., Reinecke, L., Mata, J., and Vorderer, P.

(2018). Feeling interrupted—being responsive: How

online messages relate to affect at work. Journal of

Organizational Behavior, 39(3):369–383.

Ubuntu (2020). Ubuntu documentation.

Vaske, J. J. (2011). Advantages and disadvantages of inter-

net surveys: Introduction to the special issue. Human

Dimensions of Wildlife, 16(3):149–153.

Windows (2020). Take control of your notiﬁcations in the

windows action center.

Knock Brush! Perceived Impact of Push-based Notiﬁcations on Software Developers at Home and at the Ofﬁce

221