The latest news and updates from the tech world for the business

The latest news and updates from the tech world for the business

How Developers Productivity Should Be Measured

The Discussion Continues on How Developers Productivity Should Be Measured

Productivity in software development has always been challenging to measure accurately compared to other business operations, such as sales or customer service. This stems from the intricate and collaborative nature of the software creation process, and the often murky relationship between input and the results achieved.?

However, in publication Yes, you can measure software developer productivity McKinsey argues that as is now an urgent need for it as most businesses evolve into software companies, we need a more standardized approach. Recognizing the complexity of software development – a blend of collaboration, creativity, and technicality – they propose a multi-dimensional approach to gauging productivity. This method uses both existing metrics and new ones they've developed. Their metrics consider systems, teams, and individual performance and are divided into two main categories: DORA (outcomes-focused) and SPACE (optimization-focused).?

McKinsey's added value lies in their "opportunity-focused" metrics, which target areas of potential improvement. For example, distinguishing between direct product-creating tasks like writing code (inner loop) and secondary tasks like infrastructure (outer loop) helps companies ensure developers spend more time on “value-generating activities”. This approach has been piloted in several industries with significant results, including better customer satisfaction and reduced defects. McKinsey emphasizes the importance of using metrics wisely, cautioning against oversimplified measurements that can lead to counterproductive behaviors and reinforcing the necessity of moving past outdated misconceptions about measuring software development.

McKinsey's piece became a hot topic of debate within the programming community. Numerous industry experts, like Dan North , Kent Beck , John Cutler and Gergely Orosz pointed out issues in the McKinsey's approach, highlighting, among others, an excessive focus on mere code production as the primary productivity indicator. They critique McKinsey's stance on measuring software developer productivity through quantitative metrics, arguing that such an approach oversimplifies a complex issue and misses the broader, systemic nature of productivity. McKinsey's emphasis on individual metrics overlooks the fact that productivity is deeply influenced by the entire working system.

What we learned

The debate on how to accurately measure developer productivity isn't straightforward nor concluded. It seems there isn't a one-size-fits-all solution that would satisfy both corporate business needs and individual developers' requirements. To reach a consensus, both business needs and the well-being of developers, who as we all know operate in a complex, team-oriented environment, need to be considered.


Monetizing the Software Supply Chain

Monetizing the Software Supply Chain: A Case Study of HashiCorp and Unity

Open source software, integral to modern technology from smartphones to cloud computing, is developed by a global community often working without financial incentive. This backdrop amplifies the impact of HashiCorp's decision to switch Terraform's license to the Business Source License (BSL) , permitting open visibility but limiting non-production usage. This move has spurred concerns over potential restrictions, prompting the community to create a fork, OpenTF, subsequently endorsed by the Linux Foundation as OpenTofu . OpenTofu seeks to maintain the open ethos of the Terraform ecosystem and recognize the community's crucial contributions.

This shift in licensing aligns with a broader trend among open-source vendors like MongoDB, Redis Labs, and Confluent, who are revising licenses to shield against third-party commercial exploitation, particularly by major cloud providers. The adoption of BSL by HashiCorp has sparked discussions due to ambiguities that might limit its application in competitive environments.

Terraform's value is deeply entwined with its extensive ecosystem, cultivated by community efforts, establishing it as a de-facto standard. The community will ultimately determine whether the original Terraform or OpenTofu will prevail as the industry standard.

Similarly, Unity Engine, celebrated for its royalty-free licensing, faced backlash upon introducing a "per-install" fee structure slated for January 1, 2024. The change, targeting successful developers and applied retroactively, has intensified discontent, with concerns arising around "install bombing" and the accuracy of install tracking.

What we learned

It underscores the potential challenges businesses encounter when altering models without consultation with community - in both cases, the companies misjudged the necessity of sustaining trust and open dialogue. The developer’s perspective may not always align with the vendor's business model, highlighting transparency and dialogue in such decisions even more.


Microsoft Copilot

Microsoft Goes All-In on GenAI: Launches Microsoft Copilot and Pledges to Defend Customers from Lawsuits

Microsoft has announced that it is introducing a new solution named Copilot - a new AI-powered, versatile, multitasking assistant that will be available in Windows, Bing, Edge, and Microsoft 365. Users will receive support for tasks such as coding, writing, and creating presentations. Microsoft is also adding new features to Bing, Edge, and Microsoft 365, such as personalized answers in Bing, shopping support, and chat in Microsoft 365.

Yawn, a standard announcement in 2023, right? But it has its intriguing points too - Microsoft has stated that it will protect its commercial Copilot AI tool users from lawsuits related to copyright infringement arising from content generated by these tools. The company made this decision to address potential customer concerns about the risks associated with using AI-generated content. While legal experts believe that the risk of Copilot's outputs being deemed copyright infringing is low, Microsoft decided to take extra measures to safeguard its clients.

GitHub Copilot , the first of the popular AI copilots that assist developers in coding, is at the heart of this debate. It's trained on a massive dataset of public and open-source code. Even though it's been praised for boosting developer productivity, some criticize it for potential copyright infringements. There have also been allegations about Copilot suggesting code with security flaws.

At the same time, Microsoft has introduced a public version of the Azure OpenAI service through Azure . This allows users to integrate OpenAI models, such as GPT-4, with their own private data, opening up new possibilities in data analysis and communication.

What we learned

It's evident that as the adoption and integration of AI technologies advance, providing legal and regulatory support becomes important. Microsoft's proactive approach in shielding its users from potential copyright infringement lawsuits demonstrates the importance of addressing legal uncertainties and potential risks. For wider acceptance and trust in AI solutions, especially in sectors where the line between AI-generated content and intellectual property is blurry, companies will need to ensure that users feel protected and confident in leveraging these technologies without the fear of unintended legal consequences.


Industry Cases Worthy of Examination

1. How IKEA Retail Standardizes Docker Images for Efficient Machine Learning Model Deployment

What you will learn:

About the challenges and significance of Machine Learning (ML) deployment, emphasizing the role of Docker in streamlining this process. Through IKEA Retail's? experiences, the article by Karan Honavar and Fernando Dorado Rueda showcases the synergistic benefits of Docker and Seldon-Core library in ML deployment. Practical steps for model deployment and preparation are also covered, underscoring Docker's transformative potential in the ML landscape.


2. Why We Killed Our End-to-End Test Suite

What you will learn:

About Nubank's transition from using End-to-End (E2E) testing to Consumer Driven Contract (CDC) testing for their microservices. Due to challenges like slow feedback and maintenance costs associated with E2E, they developed their own contract testing framework, "Sachem". This shift dramatically improved their deployment speed and efficiency, with the number of weekly deployments increasing significantly.



要查看或添加评论,请登录

社区洞察

其他会员也浏览了