Blog Spotlight - The Product Development Triathlon
According to Kent Beck, you can look at product development as a triathlon - it proceeds in three phases: Explore, Expand, and Extract.
Explore - the search for a viable return on a viable investment. The best strategy is to reduce the cost of experimentation and put a little investment into many, uncorrelated experiments.
Expand - the phase of rapid growth. Things break and bottlenecks appear. Processes start to be more defined gradually transforming chaotic growth until it becomes routine.
Extract - the product is now mature and the focus is on optimization, reducing costs, and maximizing the value. Changes are made with more consideration and caution.
Each phase requires different equipment, different techniques, and different training.
Understanding which phase your product is on, will allow you to tailor your approach and maximize your chances of success.
Only organizations that can execute all three phases can complete a race.
News Highlights
1. Amazon tells managers they can now fire employees who won't come into the office 3 times a week
Amazon shared updated return-to-office guidelines, requiring employees to turn up in person at least three days a week. Managers are now allowed to effectively fire employees who fail to comply with the new mandate.
In contrast, tech giants such as Nvidia and Atlassian have recently expressed their support for remote work and said they’ll stay committed to allowing employees to choose where they work from.
We expect people to be able to work from home, from a café, from an office, but we don’t really care where they do their work - what we care about is the output that they produce
-Scott Farquhar, Atlassian co-CEO
2. NASA sends a software update to spacecraft 20 billion kilometers away
After a nearly 50-year-long journey, NASA has successfully performed a critical software update that will help their space probe Voyager 2 go even farther. The update, which took almost 18 hours to complete, was transmitted to help Voyager 2 avoid the same problem that its sibling, Voyager 1, experienced last year.
Voyager 2 received the patch first to serve as a testbed for its twin. Voyager 1 is farther from Earth than any other spacecraft, making its data more valuable.
3. OpenAI’s winning streak falters with reported failure of ‘Arrakis’ project
OpenAI has reportedly discontinued the development of a new AI model named Arrakis.
Arrakis would have allowed OpenAI to run its ChatGPT chatbot more cheaply. The model, unfortunately, did not meet the necessary efficiency expectations. This decision has left some backers, including Microsoft (a 49% owner), somewhat disheartened as they had hoped to see a rapid production of large language models (LLMs).
The news was later followed by a rumor that OpenAI is in talks to sell existing employees’ shares at an $86 billion valuation, surpassing notable companies such as Stripe and China's online retailer Shein, and positioning itself as one of the world's most valuable privately held companies, trailing only behind Elon Musk's SpaceX and TikTok's parent company, ByteDance.
4. New data poisoning tool lets artists fight back against generative AI
A new tool named Nightshade has been developed to help artists fight against the unauthorized use of their artworks by AI companies for training models.
Nightshade subtly alters pixels in artworks, making them potentially disruptive to AI training, causing the resulting model to break in chaotic and unpredictable ways.
This comes in light of AI companies such as OpenAI, Meta, Google, and Stability AI are facing a slew of lawsuits from artists who claim that their copyrighted material and personal information was scraped without consent or compensation.
Sustainable Engineering Series: #2 Pinterest
Pinterest was launched in March 2010. By January 2012 they had reached 11.7 million monthly unique users, all with the efforts of just 6 engineers. Here are their key principles for sustainable growth:
It will fail. Keep it simple
Less is more
Keep it fun
Throughout 2011, Pinterest saw their user base double in size every month and a half, pushing their tech stack to its limits. Having just 3 engineers at that time, and rushing to keep the platform afloat, they over-engineered their architecture to the point of having 7 different technologies, just for data alone.
When pushed to their limits, all technologies fail in their special way
This is when their first rule was born - they decided to only choose technologies, which were mature, well-known and well-liked, and known to be stable. It makes it easier to hire talent and use their limited capacity efficiently. And you can always turn to an engaged community for help. MySQL was chosen over Cassandra and MongoDB as their main persistence layer, and sharding instead of clustering as their database scaling strategy.
To keep the cognitive load low, they chose to limit their technology and configuration options. This was initially due to vendor constraints, as AWS had a narrower range of services available at the time. Later, they intentionally kept their options limited to just a few technologies and machine configurations, which allowed them to move more quickly. Whenever they needed to scale, they simply added more of the same thing.
By February 2012 they had started to grow their engineering team quickly, reaching 40 engineers in October 2012, and so the third rule emerged. Keeping the architecture simple lets new joiners start contributing fast, focusing on creating value, rather than untangling complexity. This leads to a happier and more engaged team.