Technology to Assist Land Management: User Satisfaction with an Online Forest Management System

Abstract

Surveys of forestry professionals who actively manage, or advise upon the management of, forest lands were conducted to determine their opinions of the usefulness of a forest management decision support model. The surveys were aimed at evaluating attitudes and concerns about the eYield model, which was developed to assist in the examination of management options for eastern United States forests. The coronavirus issue that began in 2020 necessitated a virtual workshop environment to illustrate the potential usefulness of the eYield model. Pre- and post-workshop assessment surveys suggested that there was an interest by land managers in tools like eYield that are straightforward to use. The results suggested that the instructions associated with eYield were generally clearly presented, and the outcomes produced by eYield were generally representative of real-world conditions. The surveys also indicated that people represented by the sample frame were willing to consider new technology that may be used to address complex forest land management issues. Improvements suggested by survey participants may result in greater user interaction with Internet-based decision support systems that focus on the management of land.

1. Introduction

The financial aspects associated with growing forests that are necessary for private landowners to succeed and the information that is necessary to assess returns on investments in forestry have been discussed in the United States for quite some time [1,2]. Further, although agricultural activities were arguably more important in growing the economy one hundred years ago, some land areas containing forests in the eastern United States were (and still are today) the basis for successful rural economies [3]. Currently, about 39% of forest land in the United States is family-owned, and about 93% of this land encompasses 10 acres or more [4]. The most common management activities on privately owned forest lands involve the cutting and removal of trees for sale or personal use, the removal of invasive plants, and the improvement of wildlife habitat. While only about 18% of private forest landowners have sought outside advice or information regarding the management of their lands, about 81% seem interested in acquiring information about their forests [4]. The most popular sources of information include interactions with experts (nearly 40% of those seeking information) and hardcopy published works (about 55%), but interestingly, only about 25% of those seeking information indicated they used what was available through the Internet to gather advice [4]. This low amount of activity involving Internet applications might be attributed to a lack of available online tools that produce information related to both the sustainability and financial profitability of forest management decisions that small- and medium-sized forest landowners may desire [5].

Forest management decision models require growth and yield information, and perhaps the models are enhanced to allow biological or financial explorations of management options. For example, the Forest Vegetation Simulator (FVS) is a distance-independent growth and yield model that projects stand-level characteristics based on inventory data [6]. Simulated management options include intermediate and final harvests, and outcomes include several reports explaining projections that can span hundreds of years via increments of 5 to 10 years [6]. Since its initial development, FVS has been expanded to allow for the incorporation of climate change impacts into stand-level projections, along with fire and other disturbances [7]. While many growth and yield models have focused on regionally specific coniferous species grown in even-aged forest conditions, an advantage of FVS is that it accommodates mixed species, deciduous forests, and uneven-aged forest systems. Another model, the Computer-Aided Projection for Strategies in Silviculture (Capsis) software model, facilitates the growth and yield simulation of various forest types, both naturally regenerated and planted. Capsis has an open, flexible structure comprising models that focus on individual tree species and includes distance-independent tree models and stand-level simulators [8]. However, both FVS and Capsis are complex, standalone software systems that are potentially not conducive to the needs of small- and middle-sized private forest landowners. Similar models include the SIMulation and Optimization (SIMO) system, which integrates multiple models to project forest dynamics and serves as an architecture for hosting other forest simulators [9,10]. Several Internet-based forest management simulators have previously been developed [11]. For example, the 4S Tool was a decision support tool that focused on private landowners and incorporated FVS to project changes in forest condition (timber, understory plant diversity, and wildlife habitat) over a 40-year horizon [12]. To our knowledge, the system no longer has an Internet presence. Also, SIMANFOR allows a user to either use existing growth and yield models or to develop new forest growth and yield models, and while users can explore unique modeling scenarios, this system does not provide extensive financial evaluations [13]. Further, SORTIE-ND allows simulation of growth and yield, along with projections of various ecosystem services that might be derived from the management of individual stands of trees [14].

While forest simulators often focus on projecting the main components of forest dynamics, such as tree growth, mortality, and ingrowth, as well as volume, weight, and insect and disease risk [11], they often lack integration with financial considerations. Because of this, a gap seems to exist for useful and easily accessible models for non-industrial forest landowners to inform their management decisions in natural forests at the stand level. The utility of a computer application in assisting a decision-making process can be viewed positively when users suggest that the outcomes of the application are reasonable and that the application is relatively easy to navigate [15]. The objective of this study was to describe the usefulness of the models and reports that are available through a new forest management planning system (eYield). The assessment was accomplished through surveys that were conducted in conjunction with a training session devoted to the eYield management planning system. Specifically, the surveys sought to better understand what financial and biological information would be valuable to small- and medium-sized forest landowners and to those who might provide advice to these landowners (i.e., consultants), how people process information served through technology provided through an Internet-based interface, and whether the eYield model satisfactorily engages forest owners and enhances their understanding of forest potential.

2. Materials and Methods

2.1. Forest Management Planning System

Because of the gap in tools readily available to small- and medium-sized landowners with minimal data availability, the eYield model (eyield.uga.edu) was recently developed to meet these landowners’ needs. eYield is an Internet-based growth and yield model designed specifically to assist small- and middle-sized landowners in the management of their forests. The eYield model was developed through a collaboration between the University of Georgia, the University of Tennessee, and the Southern Region Extension Forester’s Office. The eYield model is a re-envisioning of the WinYield model [16], a stand-alone forest management program developed within the Tennessee Valley Authority. WinYield included a graphical user interface that allowed manual input of variables to model different management scenarios, and it incorporated growth and yield simulations with financial analyses [17]. WinYield facilitated the development of timber management recommendations at the stand level, allowing one to view alternative management scenarios of stand growth and yield estimates when combined with financial probability and produced reports that illustrated volume yields over time, transaction values, and profitability estimates. At the time of its development, typical concerns landowners had regarding the management of their property included financial profitability, standing timber inventory, and the scheduling of management activities [18].

The objectives of developing eYield included a desire to assist forest landowners in making informed management decisions, to likewise help increase management efficiency, to promote sustainability and healthy forests, and ultimately, to help ensure sustainable timber supplies. eYield focuses on natural forests, both pine and hardwood, in the eastern United States. Currently, the eYield system consists of seven natural forest simulators: loblolly pine (Pinus taeda), slash pine (P. elliottii), shortleaf pine (P. echinata), longleaf pine (P. palustris), white pine (P. strobus), upland oak–hickory (Quercus spp. and Carya spp.), and yellow-poplar (Liriodendron tulipifera). The methods for simulating natural forests are based on biometric work directly conducted in the forest systems. Two of the simulators (loblolly pine and yellow-poplar) are based on stand table projection processes, and the others are whole stand models projecting basal area and estimating volume and density from these values. The loblolly pine simulator, for example, develops a stand table (trees per unit area by diameter class) using a Weibull distribution and subsequently projects the stand table through time using growth and mortality assumptions. There are too many (over 20) references to note here that relate to the growth and yield simulation methods employed in eYield; thus, interested readers should visit the eYield tool online (eyield.uga.edu) to gain a better perspective of the modeling approaches. Validation of the biological (growth and yield) and financial outcomes from eYield was accomplished through direct comparison with outcomes from WinYield (using a vintage 1995 laptop that could operate the WinYield program). Several management scenarios focusing on each of the seven simulators were compared to ensure fidelity between the eYield and WinYield modeling outcomes. As with other decision support tools [14], usability principles of efficiency, effectiveness, and satisfaction were considered during the design phase of eYield.

Like WinYield, inputs to the eYield system include stand size, site index1, log rule2, duration of tree growing activity, forest density, average tree age, and potential harvesting activities. Additionally, eYield requires information on tax rates, management costs, and the length of the planning horizon. Several reports are created, including a hazard rating related to infestations of southern pine beetles (available only for pine models), cashflows by transaction, cashflows by year, financial profitability, growth and harvest metrics, market conversions, and a woodflow summary. These reports are offered to the user as downloadable PDF files. In terms of functionality, a user of eYield would provide the model with some basic information on the conditions of a stand of trees (size, site index, age, and perhaps density of trees). At least one harvest (thinning or final) would need to be planned within a period of time defined by a time horizon. In conducting this analysis, the user needs to provide an estimate of their personal tax rates and the potential value (per unit) of harvested wood products. The reports generated by eYield provide the user with a glimpse of the biological growth and financial outcome potential of managing the stand. The designation of additional harvests and the ability to acknowledge a variety of management actions within the time horizon provide the user with the ability to estimate before-tax and after-tax net present value and other financial aspects of managing the stand. A series of alternative management plans can be produced with minor adjustments to the financial and management action assumptions that are provided, allowing the user to compare alternative futures for the stand.

The major difference between WinYield and eYield is that eYield is available through the Internet, making it readily accessible to anyone with an Internet connection. It has been developed in a responsive manner so that it properly functions on a computer, tablet, or smartphone. While WinYield was developed in FORTRAN, eYield has been developed using the Python programming language, and the front-facing Internet interface has been created in a WordPress environment.

2.2. Surveys of User Satisfaction of the Forest Management Planning System

Workshops were designed to illustrate the function of the forest management planning system. Potential workshop participants were recruited using publicly available, valid, and unique e-mail addresses associated with forest management professionals working in the State of Tennessee (USA). Sources of information on potential workshop participants included the Tennessee Division of Forestry, the Tennessee Wildlife Resources Agency, the U.S. Forest Service, the University of Tennessee, the Tennessee Forestry Association, the Association of Consulting Foresters, the Society of American Foresters, and personal contacts of the authors. From this effort, a total of 99 potential workshop participants represented the sample frame.

One main issue in recruiting workshop participants was that the workshops were conducted during the height of the COVID-19 pandemic, which resulted in lower attendance than anticipated. Due to COVID-19 pandemic guidelines issued by the U.S. Centers for Disease Control and Prevention, the original plan of having in-person workshops was shifted to the use of online workshops that were conducted through the Zoom conferencing platform. Potential participants were sent an e-mail message containing a short synopsis of the eYield system and how it may apply to their work processes, information about the workshop meetings, and links to the remotely accessed meeting rooms. Between 20 October and 18 November 2020, five workshops were offered. Each workshop lasted approximately 2 h. The workshops involved guiding participants through eYield system functions and the website interface, and instructors acted carefully to avoid remote instruction issues that have been noted recently by others [19,20]. Participants were introduced to the different simulators available and were allowed to input their own management scenario assumptions into eYield.

Pre- and post-workshop assessment surveys were conducted (Appendix A). Each survey was reviewed by the Institutional Review Board at the University of Tennessee, which determined that the surveys did not meet the criteria for requiring human subject approval. The identity of survey respondents was kept anonymous, and participation in the surveys was voluntary. Due to the ongoing COVID-19 pandemic, surveys were provided to workshop participants online instead of in person. As participants were attending workshops remotely, a link to the pre-assessment survey was provided using the chat function in Zoom before the start of each workshop. Upon the completion of the workshop, a link to the post-assessment survey was again distributed to workshop participants using the chat function in Zoom. Surveys were created using QuestionPro (https://www.questionpro.com/ accessed on 19 October 2020) survey software. The surveys were composed of multiple-choice, Likert-scale ratings, and open-ended questions. This format has been used in recent surveys designed to assess the perceptions of small groups of people about the usability of decision support tools [14].

The pre- and post-workshop assessment surveys included questions concerning the desirability of outcomes. Survey respondents were provided a list and description of reports generated by eYield: cashflow by transaction, cashflow by year, financial profitability, market conversion, bark beetle, growth and harvest, and woodflow summary. Survey respondents were then asked to select which of these reports they might find useful, using a Likert scale ranging from not very useful to very useful. Respondents in both surveys were asked how long they have used Internet- or computer-based models to aid in their land management objectives. Relatedly, survey respondents were asked to indicate whether they preferred information presented digitally or in hardcopy format. Additionally, they were asked if they had ever participated in remote or computerized training. If respondents indicated that they had previously participated in remote or computerized training, they were asked to rank their experience with online learning on a scale from being the worst way to learn (1) to the best way to learn (5). Finally, survey respondents were asked to indicate whether they were a forestry or land management professional.

In only the post-workshop assessment survey, respondents were asked several additional questions specific to their experience using eYield. First, respondents were asked for their opinions on the forest simulation model results compared to their expectations. If they indicated that the results did not meet their expectations, they were provided an opportunity to explain which parts of the results veered from the results they expected. Respondents were also asked to provide examples of potential changes or features that would be necessary before they would use the eYield model in practice. Survey respondents were offered an opportunity to provide any additional comments, thoughts, or opinions about eYield or the workshop.

Where possible, pre- and post-assessment survey results were compared using Pearson’s chi-squared (X²) test and the non-parametric Wilcoxon signed-rank test. The Wilcoxon test was chosen due to the non-normal distribution of the results as well as the small sample size of survey responses. All statistical analyses were conducted using RStudio. In the post-assessment survey, Likert-scale rankings were converted from qualitative responses (i.e., very useful) to quantitative values ranging from −2 to 2. This goal attainment scale is based on five discrete levels, ranging from indications that a component of the eYield model is very useless (−2), indifferent (0), or very useful (2). This type of scale was originally designed for assessments in the field of medicine [21], where a value of 0 is used to indicate a neutral opinion on a subject [22]. Further, using Likert-scale ranking responses, Vaske’s Potential for Conflict Index (PCI) was used to visually highlight where survey respondents’ opinions of different components of eYield converge or diverge [23]. PCI has been used to graphically compare consensus and conflict through a single value that integrates the central tendency, dispersion, and distribution of responses to survey questions [24]. PCI values range from 0 to 1, with 0 indicating no conflict or complete consensus, 1 indicating great conflict or polarization, and values falling in the middle of this range indicating a neutral or a lack of agreement or disagreement on a topic [25]. An associated bubble graph allows one to visualize the PCI findings, where the size of a bubble suggests potential for conflict (small bubbles have less potential and large bubbles have more potential), and the center position of a bubble represents the mean response [26]. Here, we employ a −2 to 2 scale in the graphic to account for acceptable (above 0) and unacceptable (below 0) responses to the values of agreement of the survey respondents. Other relatively similar scales have been used to represent PCI values [24,27].

3. Results

The workshops were attended by 55 people, and this subset of the larger sample frame solicited for participation (n = 99) represents a 55.5% participation rate. Given the timing of the workshops during the COVID-19 outbreak, this participation rate seems relatively good. Of those who attended a workshop, 28 (51%) responded to the pre-workshop assessment survey, while 19 (35%) responded to the post-workshop assessment survey. Nearly all respondents identified as forestry professionals in the pre- and post-workshop surveys, 96% and 100%, respectively, yet the retention of survey respondents from the pre- to the post-workshop survey was a concern. The workshop participation rate was higher than similar recent surveys of foresters in the United States [28], yet the survey response rate was lower.

In both the pre- and post-workshop assessment surveys, respondents were asked to identify which terms they were familiar with associated with components (reports, etc.) of the eYield model to determine if users were better informed about concepts associated with the growth and yield modeling effort (

Table 1. A list of reports available through eYield and a brief description of each.

Report	Description
Cashflow by transaction	Type, amount, and taxes associated with each income (revenue) and expense transaction
Cashflow by year	Aggregate before- and after-tax revenues and expenses
Financial profitability	Measures of the financial plan, including cost-benefit, net worth, and rate of return
Market conversion	Product dimensions and applicable prices at the time of harvest, including stand harvest statistics, wood volumes and weights, and product prices
Bark beetle	Hazard rating report for southern pine beetle
Growth and harvest	Pre- and post-harvest stand statistics
Woodflow summary	Standing and harvested stand statistics and marketable wood volumes

) after working with eYield. On average, respondents were familiar with 3.48 terms according to the pre-workshop assessment survey, while they were familiar with 4.67 terms in the post-workshop assessment survey. This suggests that an improvement in understanding regarding the language used within the modeling system was evident. Using a significance level of 95% (p = 0.05), a comparison of the pre- and post-assessment responses resulted in a X² of 0.53, indicating a non-significant difference between the two samples. Respondents also did not have statistically different opinions on the usefulness of the various reports and simulation options, according to the Wilcoxon signed-rank test. This result is likely a function of the small sample size available for analysis. The average ranking for the usefulness of the biological simulation components (growth and harvest and woodflow summary) was 0.33 (slightly above neutral) compared to the average ranking value of 0.04 (neutral) for the usefulness of the financial simulation components (cashflow by transaction, cashflow by year, financial profitability, and market conversion). This suggests that the growth of trees and associated potential harvest outcomes seemed to be of more value to the survey participants than the financial analyses that eYield could provide.

The similarities in views of respondents for both biological and financial (Figure 1) reports from eYield were visualized using the PCI graphic. The resulting PCI values indicated a general indifference by users to the eYield reports in the post-assessment survey, as most PCI values were just under or just over 0.5, suggesting a lack of agreement or disagreement with reports. However, respondents’ thoughts regarding (a) the financial profitability report and (b) the growth and harvest report edged slightly towards conflict and polarization, as these PCI values were 0.722 and 0.611, respectively. Using the size of the circles created in the PCI graphic, a degree of variability in responses can be visualized. The center position of each bubble also represents the mean response, so while it may seem odd that the PCI values do not reflect the position of each bubble with respect to the y-axis, the PCI values also capture variability in agreement on the topic (in this case, the value of each eYield report).

In the pre-workshop assessment survey, a majority of the respondents (82%) indicated that they had used computational or Internet-based land management resources for 6 years or more. In the post-workshop assessment survey, there was a decrease (74%) of those respondents that indicated they had used these resources for 6 or more years, likely due to a reduction in the number of respondents that completed the post-workshop assessment survey. This suggests that 20 to 25% of survey respondents use other means, or none at all, for assessing the potential outcomes of forest management actions. Additionally, due to the nature of the workshop occurring online through Zoom, implicit bias may be present, with respondent opinions skewed toward those with some level of comfort with computer technology and previous experience working within an online environment. Ninety-two percent of pre-workshop assessment survey respondents indicated that they had participated in a minimum of one online or remote training course. Using a Likert scale ranging from 1 (unfavorable) to 5 (favorable), respondents rated the online training experiences at 3.4, on average. In the post-workshop assessment, respondents were asked to rank their experience in the eYield workshop compared to their experiences with other remote or computer-based training and ranked the eYield training experience as 3.5, on average. These observations suggest that the online training experience, particularly in this case with respect to new technology, was neither favorable nor unfavorable and may suggest that in-person training may be more effective in this regard, although this assertion was not tested.

Interestingly, in the pre-workshop assessment survey, 68% of survey respondents noted that they would prefer to receive information digitally, while 32% preferred hardcopy (or paper). Following the completion of the workshop session, 74% of respondents preferred digital information, compared to 26% preferring information via paper. Again, the nature of how the workshop was presented over Zoom may have biased these results, as some respondents may have become adjusted to using this technology for longer periods of time. Of those survey respondents that preferred receiving information on paper, 88% indicated that they had used computers during their work activities for 6 or more years in the pre-workshop assessment survey, while 80% indicated the use of computers for 6 or more years in the post-workshop assessment survey. Focusing on the questions only asked in the post-assessment survey, survey respondents were asked if they found any of the instructions related to the eYield interface to be unclear. Sixty-eight percent of respondents indicated they did not find any instructions to be unclear, while 32% indicated that at some point in using eYield they did find instructions to be unclear. The structure of the online training sessions may have prevented a more open, face-to-face discussion of unclear instructions, but again, this comparison was not tested.

Survey respondents were asked if they believed the model results were comparable to their expectations based on their real-world experience. Of those that responded, 74% felt that the model results met their expectations, while 26% did not. When offered a prompt to comment on what portion of the eYield report results did not align with expectations, survey respondents indicated that the difference in their expectations may have been the result of guessing at model simulation input values, which may have resulted in revenue results that seemed high. Further, the survey respondents noted that financial forecasting for mixed forest stands was limited by an inability to input current market prices on a tree species-specific basis and that the eYield system seemed most useful for landowners with more homogenous stands of trees. On a positive note, 81% of post-workshop assessment survey respondents indicated that they would use eYield in the future.

4. Discussion

The management of small and medium-sized forests in the eastern United States is important to both the landowner and the nearby forest product markets. We conjecture that forest landowners could benefit from information about the economic aspects of their land management strategy, as estimates suggest that only about 11% of the forest landowners in this region have sought advice or information on the management of their forests [30]. Integrated models that consider forest productivity along with financial considerations (rate of return, cash flow timing, and tax considerations) may therefore be of value in assisting forest landowners with their analysis of forest management strategies [5]. This information could be provided by tools such as eYield that are designed to assist forest landowners in understanding potential forest management options.

This study aimed to understand the usefulness of a forest management decision model that was developed for the natural forest conditions of the eastern United States. While our sample frame was consistent with recent work in this area [14,31], the relatively low response rate of the two surveys can be a concern when making direct inferences about the functionality and usability of eYield and limits the statistical validity. However, others have utilized small groups of survey respondents to successfully gather opinions of software functionality and end-user satisfaction [32]. And although it is acknowledged that each survey may be provided with a unique sample frame, others have shown that reasonably reliable information may be obtained from a survey sample size as small as about 14 people [33]. Therefore, the feedback we obtained through the pre- and post-workshop assessments of eYield should be of value in addressing future improvements to the model.

The insights obtained from practicing professionals are important, as software application (system) quality, along with outreach and training (service) quality, significantly affects end-user satisfaction with a computer application [34]. Further, interest in and acceptance of a decision support system may be affected by the construction and user-friendly nature of the software employed [35]. The classical concept of usability involves effectiveness, efficiency, and satisfaction, the latter of which is subjective (perceived), while the former two are fairly objective [36]. For these reasons, landowners and practitioners can provide valuable insight into management problems that might not otherwise be uncovered, insight that may form the basis of integrated approaches to land management [35]. User acquisition and retention, therefore, play a pivotal role in the success of any computer application. Persistent use of an application may lead to improvements in management productivity; ease of access, ease of learning, and usability are key factors affecting persistent use [37]. Since the time of the workshops, we have addressed some issues highlighted by workshop attendees. Specifically, the speed of the eYield model has been greatly improved by moving it from a server managed by the University of Georgia to an Amazon Web Services (AWS) cloud server. We have also created video-based guides through the simulation process and a user guide with thorough descriptions of the growth and yield models used, as others have suggested extensive documentation is important in ensuring adoption of software products [38] and that the Internet may be the most effective means of communication about new ideas [39].

Analyzing feedback from survey respondents using the PCI is an interesting way in which the consensus or polarization of ideas can be viewed. PCI was initially developed as a tool to communicate the distribution of variable responses to non-technical audiences, where statistical training was neither necessary nor commonly held by people in those audiences [27]. As our results suggest, there is a general indifference amongst survey respondents to the reports generated by eYield, although two eYield reports edged towards conflict and polarization opinions amongst survey respondents. However, the bubbles employed in the PCI graphic straddle the indifference line, indicating a minimal level of consensus [27], while some difference in opinion regarding the value of these reports may exist.

Overall, this study contributes toward the advancement of the success of decision support systems designed for land management and offers insights into improvements that could be made to improve end-user satisfaction with computer applications. As others have suggested, the benefits of sophisticated design choices for a computer application may not be viewed similarly by different people, and, thus, opinions concerning the usefulness of a model will vary amongst the target audience [40]. Computer applications that do not perform well in individual cases suggest that human experience cannot be entirely replaced by computer models [41]. Yet the results we obtained from the surveys conducted also highlight the continued need for outreach and training. As eYield is designed for small- to medium-sized landowners, future opportunities may exist for training and additional feedback on the eYield model by targeting these ownership groups, actions that may further encourage wider adoption of the model [34].

5. Conclusions

The objective of our study was to evaluate whether an object (a forest management planning model) might actually be used by the intended audience (foresters) and to understand their concerns about modeling effectiveness, efficiency, and satisfaction with respect to the user experience. We postulate that with a broader acceptance of the eYield model as a relatively easy-to-use tool, more landowners and foresters could use it to increase the efficiency of land use. A focus on ease of use provides a frictionless environment between the user and the program. The program should operate in the way that the programmers intend and work in a way that is intuitive to the user with minimal effort. The surveys were an attempt to evaluate users’ abilities, attitudes, and thoughts about the eYield model. Survey respondents indicated a willingness to accept new technology to address questions that are environmentally complex and highly variable in association with future forest growth on the lands that they own or manage. There were some common reactions to the usefulness of eYield reports, although the post-workshop survey indicated a slight disagreement regarding the usefulness of two reports. The reasoning people use and the processes they employ to develop insight into the usefulness of the modeling platform and the outcomes potentially generated can inform further beneficial development of services such as eYield. These observations and judgments of the survey respondents may help provide private forest landowners, who may not have the wherewithal to otherwise employ a consultant, with more effective alternatives for assessing different courses of action for their land. Ultimately, the benefits derived from an investment in training need to clearly relate to improvements in management productivity and efficiency for new technology such as eYield to be of value.