The E/Jugaad Manifesto
Shrink Accelerationism Timelines and Budget by Embracing Imperfection
This post is a followup to Effective Jugaad: An Ideology for Navigating Complexity and Uncertainty in the 21st Century. Go check that out first if you are middle of the bell curve or not Brown (large overlap between the two groups). Whole idea is inspired from this Twitter exchange.
At its core, jugaad is a form of frugal innovation that uses improvisation, creativity, and collaboration to achieve objectives in an efficient and cost-effective manner. It is a mindset that is characterized by a make-do attitude and a willingness to work with whatever is available. You can make Jugaad effective by solving problems for multiple people at once instead of just yourself.
I’m not going to waste your time by explaining the origin story of this whole ideology first. We will straightway dive into the principles of e/Jugaad which will later be followed with the elaboration and maybe some examples.
E/Jugaad Leadership Principles
Usability is your first and only priority
Perfectionism is Evil
Cheap is Better
Time to market > Time to Build
Best Practice Guide is whatever works
Copying is OK, Stealing is Not
Iterations > Research
Usability is your first and only priority
You should create products/solution which work IRL. Publishing a github repo with 10 shoddy scripts is not fine. Write another bash script which calls those 10 scripts in the correct order is fine. Aim to be Amazon Echo with 1000 third party integrations and not Google Nest device.
Perfectionism is Evil
Perfectionism is the root cause of all delays. It is also the cause of redundant regulations plaguing the society. People are smart (including > 3yo kids). As long as your product works for your particular problem it is fine. Add another jugaad for edge-cases when you encounter it.
Eventual Consistency in databases with Quorum writes is a good example of this.
Cheap is Better
Carbon-Credits was a ZIRP era phenomenon. Just use the cheapest solution that is available. You can pass that cost-saving to the customers OR use it to increase scale and buy those solutions even cheaper.
Hyundai Cars providing a no-airbags options during purchase in India is a good example. Trust your user!
Time to market > Time to Build
The shoddy generator you created using 10000 bicycles dynamos can be used to power handlooms in villages. Don’t keep it to yourself. Feel no shame in publishing imperfect or borderline degenerate products. Just look at Tensorflow’s code.
Best Practice Guide is whatever works
Derivative of the second principle banning perfectionism. Best practice guides are written by interns at big companies which are not actually followed anywhere and only cause delays.
Change your model learning rate based on a divine inspiration from the universe on each run to get better results. Don’t believe me. Check out ‘Prememptive Plan’ on page 21 of this OPT175B Training log.
Copying is OK, Stealing is Not
Someone else launched a solution to the problem you were working on and actually performed well in the market. Drop everything and Just copy it.
Tensorflow would have been successful if they simply adopted this approach. Microsoft did with Azure and now look where they are. Stealing is not fine unless the you steal their labour and compensate them probably.
Iterations > Research
Research papers are mostly a sham. Most research is not reproducible OR will take 20 years to work IRL scenarios. You should aim to create a half-baked product inspired from random research papers and push them to market if it helps a person.
You can always improve your product in next iteration. Reduce the time between multiple iterations rather than attempting to tackle every problem in the same iteration. Move Fast, Break Things is inspired from this philosophy.
Conclusion
With this we welcome you to e/Jugaad samiti. You will henceforth focus on building usable solutions which may work at scale. You will not be deterred by the impediment during scaleup and simply use more Jugaads (copied from your competitors) to evolve your product.
For any support, you can always consult Telt who is my superior, full-time shitpoaster and part-time employee to fund brown chat’s GPU clusters in Alibaba cloud (since it is cheaper).
Additional References