[docs] Update the ROADMAP

ROADMAP.md

@@ -2,14 +2,63 @@
 
 <p class="description">The roadmap is a living document, and it is likely that priorities will change, but the list below should give some indication of our plans for the next major release, and for the future.</p>
 
-Our priorities:
-1. **Material Design**. The Google Design team has recently released a new iteration on the Material Design specification which is focused on *themability*. This is great news! We will do our best to keep up to date. It also validates that heavily investing in customizability for v1 was a good call. We will continue our efforts in this area.
-1. **More utils**. We want to work on new utility components: Spacing, Display, Color, etc.
-1. **Performance**. We can’t optimize something we can’t measure, and we don’t currently have a CI performance benchmark, so we will need to build one and start investigating bottlenecks.
-1. **Learning materials**. The quality of the documentation is equally as important as the quality of the implementation, and while the reference documentation is comprehensive, we could author a learning tutorial like Next.js has done, or an egghead.io course.
-1. **Themes**. We want to provide common layouts example to make getting started even easier. We also plan on adding more premium themes.
-1. **Documentation**. We want to translate the documentation into Chinese and other languages. Any help is welcomed!
-1. **Bundle size**. The library needs to be as small as possible, so we need to work on solutions to further reduce the bundle size, for example, supporting Preact or Nerv can help.
-1. **Type checking**. We will continue to improve the TypeScript and possibly the Flow coverage of the library.
-
-You can follow [our milestones](https://github.com/mui-org/material-ui/milestones) for a more detailed overview.
+## Methodology
+
+We work on the problems that resonate the most with our users.
+Please **upvote** 👍 the issues you are the most interested in on GitHub.
+Thank you for participating [in our developer survey](https://material-ui.com/blog/2019-developer-survey-results/).
+
+## Our priorities
+
+Here are our top priorities, ordered by importance:
+
+- 1.0 - **More components**. 🧰 People want more components. This is challenging for us to handle. Building and maintaining a professional component takes an incredible amount of time.
+Everybody underestimates it, we have a relatively small core team. We are going to try the following strategy:
+  - Identify frequently needed components. We have different leverages for this: the developer survey answers, the GitHub upvotes, the Algolia search volume, the Google search volume, documentation usage volume, npm downloads, etc.
+  - Prioritize the creation of frequently needed components.
+  - Optimize for the people maintaining the components. The more people maintaining the components the better.
+  - Encourage the usage of third-party components if they already exist and are well maintained. We can create [a search experience](https://xpvrpr7r4o.codesandbox.io/) for this.
+- 0.5 - **Better customization.** 💅 We want to make our component customization intuitive, no matter if you are using global CSS or styled-components:
+  - Improve the documentation.
+  - Better support for styled-components.
+  - Work on the core solution.
+  - Provide [more examples like this](https://mui-treasury.com/components/card).
+- 0.3 - **Better documentation.** 📚 It's a wide topic. We don't know exactly what people want. We will focus our effort in the following areas:
+  - Beginner tutorials & Video lessons.
+  - Page Layout Examples. They get people started really quickly, we need more of them!
+  - Page documentation rating 🥇🥈🥉. We will integrate a rating module in all our documentation pages. This way, we can collect high-quality data points and prioritize the pages that need the most to be improved.
+  - Internationalization. Not everybody can read & understand English. The documentation pages are synced with [Crowdin](https://translate.material-ui.com/). They should always be up-to-date.
+- 0.3 - **TypeScript.** 📏 There are two dimensions to this problem:
+  - The documentation. We want to provide a TypeScript variant to most of our demos.
+    It should make it straightforward to use Material-UI with TypeScript. You can help us out in [#14897](https://github.com/mui-org/material-ui/issues/14897).
+  - The definitions. We are **continuously** improving them. We are aware that correctly typing withStyles() is cumbersome. We encourage people to use [makeStyles()](/css-in-js/basics/#hook-api). It should be much better with this Hook API.
+  The codebase is written in JavaScript, we don't plan on migrating it to TypeScript in the near future. Maybe for v5, we will see.
+- 0.2 - **Performance.** 🚀 React abstraction has a cost. The more components you render, the slower your page will be. We can notice stark differences when rendering a large table or list.
+Performance is all about doing less work. We have the following leverages:
+  - Make the core faster. Using our table components is x3 slower than using native table elements.
+    What's the difference? a Material-UI table component creates many **intermediary components**, slowing the rendering.
+    We are working on removing these intermediary components. We are migrating the codebase from class based components to hook based components.
+    [It will yield a theoretical **+30%** performance gain](https://github.com/mui-org/material-ui/issues/10778#issuecomment-472840548), as well as making the DX much better.
+    Here is the React Dev Tools output for the same node:
+
+    **Before (v3)**
+
+    ![Before](https://pbs.twimg.com/media/D1obANqX4AAAZgJ?format=png&name=360x360)
+
+    **After (v4)**
+
+    ![After](https://pbs.twimg.com/media/D1oZvtsXcAIyAb4?format=png&name=360x360)
+  - Avoid re-rendering. It's the responsibility of our users to prune the React rendering tree efficiently.
+    Most of our APIs are too low level to implement efficient memoization (React.useMemo, React.PureComponent, React.memo).
+    If you find a good opportunity for it, let us know, we will be happy to work on the problem.
+  - Avoid rendering. We are documenting [how to implement virtualization](/demos/tables/#virtualized-table) for the Table components. It's important to consider it after 100 items.
+- 0.2 - **Bundle size.** 📦 You can keep track of our progress [following bundlephobia.com report](https://bundlephobia.com/result?p=@material-ui/[email protected]).
+It's a continuous effort. v4 is 15% smaller so far, while adding new features.
+We are eager to find new bundle size reduction opportunities.
+  - JSS. Can we reduce the overhead of this module?
+  - Migrating from class components to hooks [helps](https://twitter.com/olivtassinari/status/1106905745264652289).
+  - Popper.js is working [on a smaller v2](https://bundlephobia.com/[email protected]) version.
+  - We can replace [react-event-listener](https://bundlephobia.com/[email protected]) with the hook API.
+  - react-transition-group will remove [react-lifecycles-compat](https://bundlephobia.com/[email protected]).
+- 0.2 - **Material Design Update.** 🎀 The material design specification is evolving, so should we. We have a few open issues about it.
+- 0.1 - **Accessibility.** ♿️ We have relatively few accessibility issues, but we are eager to address them all.

docs/src/pages/demos/tables/tables.md

@@ -70,7 +70,9 @@ A simple example with spanning rows & columns.
 
 ## Virtualized Table
 
-In the following example, we demonstrate how to use [react-virtualized](https://github.com/bvaughn/react-virtualized) with the `Table` component. It renders 200 rows and can easily handle more.
+In the following example, we demonstrate how to use [react-virtualized](https://github.com/bvaughn/react-virtualized) with the `Table` component.
+It renders 200 rows and can easily handle more.
+Virtualization helps with performance issues.
 
 {{"demo": "pages/demos/tables/ReactVirtualizedTable.js"}}