+1

Curiosity.md

reviewed

··· 5 5 - Never stop [[learning]], practice [[Thinking|rational thinking]] and internalize [[mental models]] and other concepts that will help you think more effectively. 6 6 - Knowledge is a powerful tool. The more you [feel like a noob](http://paulgraham.com/noob.html), the better. Feeling stupid now is better than feeling stupid in 10 years. 7 7 - [Reality has a surprising amount of detail](http://johnsalvatier.org/blog/2017/reality-has-a-surprising-amount-of-detail). Knowing more about the world makes you enjoy it more. [Understanding how music is made increases the pleasure you get from music](https://youtu.be/JbVfcZxfIZo). 8 8 + - Real things exist in essentially infinite resolution. [Looking closer always reveals more, and it’s often not what you’d expect](https://www.raptitude.com/2023/10/the-truth-is-always-made-of-details/). 8 9 - Do stuff! Whatever is you work on, is worthwhile as long as you share your learnings. In the worst case, if your [[ideas]] don't work out, the community will learn why that approach doesn't make sense. 9 10 - Remix ideas. Ideas are impacted by [[evolution]]. The most useful ones survive and evolve. [Innovation is product of the combinations of ideas](https://youtu.be/XUAIIQFoufs). Everything is a remix! 10 11 - Fail early and often. There is only one guaranteed way you'll won't get something you want, and that's not to pursue it. [Mistakes](https://meta.wikimedia.org/wiki/So_you%27ve_made_a_mistake_and_it%27s_public...) are the portals of discovery.

+3 -3

Datathons.md

reviewed

··· 6 6 2. Do a basic data exploration. Try to understand the problem and gather a sense of what can be important. 7 7 3. Get baseline model working. 8 8 4. Create `scikit-learn` compatible metric if needed. 9 9 - 5. Try different approaches for preprocessing (encodings, Deep Feature Synthesis, lags, aggregations, imputers, ...). If yopu're working as a group, split preprocessing feature generation between files. 9 9 + 5. Try different approaches for preprocessing (encodings, Deep Feature Synthesis, lags, aggregations, imputers, ...). If you're working as a group, split preprocessing feature generation between files. 10 10 6. Plot learning curves ([sklearn](https://scikit-learn.org/stable/modules/learning_curve.html) or [external tools](https://github.com/reiinakano/scikit-plot)) to avoid overfitting. 11 11 7. Tune hyper-parameters once you've settled on an specific approach. ([optuna](https://optuna.readthedocs.io/)). 12 12 8. Plot and visualize the predictions (histograms, random prediction, ...) to make sure they're doing as expected. Explain the predictions with [SHAP](https://github.com/slundberg/shap). ··· 37 37 - [Quick Tutorials](https://www.kaggle.com/c/jane-street-market-prediction/discussion/198951) 38 38 - [Tsfresh](https://tsfresh.readthedocs.io/en/latest/) 39 39 - [Fold](https://github.com/dream-faster/fold) 40 40 - - [Prophet](https://facebook.github.io/prophet/docs/quick_start.html) 40 40 + - [Neural Prophet](https://neuralprophet.com/) 41 41 - [Darts](https://github.com/unit8co/darts) 42 42 - [Pytimetk](https://github.com/business-science/pytimetk) 43 43 - - [Sktime](https://github.com/alan-turing-institute/sktime) 43 43 + - [Sktime](https://github.com/alan-turing-institute/sktime) / [Aeon](https://github.com/aeon-toolkit/aeon) 44 44 - [Awesome Collection](https://github.com/MaxBenChrist/awesome_time_series_in_python) 45 45 - [Video with great ideas](https://www.youtube.com/watch?v=9QtL7m3YS9I)

+3

Decentralized Protocols.md

reviewed

··· 20 20 - A decentralized protocol can work with a centralized provider. It has the benefits of both (might be fast but no lock users in). 21 21 - A major downside of decentralized protocols/networks is that they tend to perform poorly. Hubs are efficient. 22 22 - [It's the properties decentralization gives us that we care about, not decentralization itself](https://haseebq.com/why-decentralization-isnt-as-important-as-you-think/). Decentralization is a global, emergent property. You can feel latency, you can feel transaction fees, but networks ostensibly feel the same whether they’re centralized or decentralized. Decentralization is valuable when it lets you do new things fundamentally better, not old things fundamentally worse. 23 23 + - If a system requires a centralized part, a great alternative is give the user the ability to point to other centralized things taking care of that part. 24 24 + - If you have a protocol, try enforcing the desired behavior using the protocol. Your ideas of how to solve it might not be the best and adding a protocol restriction (incentives/penalties) will make people figure out. 25 25 + - When building a technology, consider: [does this centralize or decentralize power?](https://geohot.github.io/blog/jekyll/update/2021/01/18/technology-without-industry.html)

+11 -6

Talking.md

reviewed

··· 11 11 - Ask for help instead of demanding it. 12 12 - [Say "yes, and..." (accept ideas and build on top of them). Make other people feel good. Be positive (optimistic language)](https://youtu.be/VhkcmN-CCYw). 13 13 - Favor interrogative-led questions over leading questions. A leading question attempts to get the [[Listening|listener]] to agree or disagree with a premise you feed to them. An interrogative-led question often begins with the words: who; where; what; when; why. "Did you like the movie?" vs "What did you think about the movie?". 14 14 - - Make [interesting questions](https://docs.google.com/document/d/1K_iFxFt9lh1i0mxKRIhOSd2e8X1LNvxnihbChKXhyOc/mobilebasic) to know people better. Use FORD (Family, Occupation, Recreation, Dreams) as fallback topics. 15 15 - - What's the change you'd do to the current political system? 16 16 - - Why hasn't happened? 17 17 - - [[Thought Experiments]] 18 18 - - Will you eat synthetic meat? 19 19 - - [Things to argue about over the holidays instead of politics](https://dynomight.net/arguments/). 20 14 - Remember people's names. Humans want to be significant. We strive for meaning. 21 15 - Dishonest Flattery while effective in some cases, will destroy you long term. If you are unable to find a compliment about another human being that is truthful, you're not trying hard enough. 22 16 - In difficult conversations, keep in mind the ultimate purpose and [[Mindfulness|be calm]]. You don't need to win the conversation. Ensure safety. Control your emotions. 23 17 - Share your facts as stories, not global facts. Logic doesn’t persuade people. Clarity, storytelling, and appealing to self-interest do. 18 18 + 19 19 + ### [Interesting Questions](https://docs.google.com/document/d/1K_iFxFt9lh1i0mxKRIhOSd2e8X1LNvxnihbChKXhyOc/mobilebasic) 20 20 + 21 21 + - Use FORD (Family, Occupation, Recreation, Dreams) as fallback topics. 22 22 + - What's the change you'd do to the current political system? 23 23 + - Why hasn't happened? 24 24 + - [[Thought Experiments]] 25 25 + - Will you eat synthetic meat? 26 26 + - [Things to argue about over the holidays instead of politics](https://dynomight.net/arguments/). 27 27 + - Cognitive reflection test 28 28 + - What is your life default stress/drama level 1 to 10? Where are you now?

+1 -1

Thinking.md

reviewed

··· 24 24 - Avoiding stupidity is easier than seeking brilliance. Think backward so that you can avoid failures. 25 25 - Research before judging! We do not know what we don't know. Gather as much context as you can before making any final statement. 26 26 - [Absolute truth is relative and everyone is doing the best they can](https://letterstoanewdeveloper.com/2019/08/12/there-are-no-adults-in-the-room/). These are opportunities for you to help and learn more about the world. 27 27 - - Think in distributions instead of [magic answers](http://cassandraxia.com/cogbiases). The world is [analog and not digital](https://waitbutwhy.com/2019/12/political-disney-world.html), continuous and not discrete. Nuance is everywhere. 27 27 + - Think in distributions instead of [magic answers](http://cassandraxia.com/cogbiases). The world is [analog and not digital](https://waitbutwhy.com/2019/12/political-disney-world.html), continuous and not discrete. [Nuance is everywhere](https://www.raptitude.com/2023/10/the-truth-is-always-made-of-details/). 28 28 - Real people are complex and flawed, [full of faults and biases](https://upload.wikimedia.org/wikipedia/commons/6/65/Cognitive_bias_codex_en.svg). Each turn of events is mired in potential positives and potential negatives, which is a mess to sort out. 29 29 - [Fundamental Attribution Error](http://www.aaronsw.com/weblog/nummi): we attribute people's behavior to their personality, not their situation. 30 30 - Digitizing an analog view will result in some loss of information. In that world, everything is good or bad, everyone is smart or ignorant, ones and zeros. Mistrust simple comparisons.