Doing research in MLC is not too different from doing research elsewhere. You need a few key components to get started: an idea (e.g. *"I want to stabilize GAN training"*), any number of collaborators (can be zero), a regimen (e.g. *"working on it X hours a week"*) and a goal (e.g. *"publishing it at a major conference in a year"*). And all of these should be established upfront and revised and refined iteratively.

But one key difference of doing research in MLC — or any non-traditional, non-institutionalized setting — is that none of the above *come as a given*. While in a grad school lab, or an employment-based research lab, you may be handed an idea, automatically assigned to a team, a manager and mentor, naturally following a regimen imposed by exams, qualifiers, perf reviews and/or just peer pressure, and have a clear goal to graduate or to be promoted, here, you will have to act proactively on your own to gather all those ingredients onto your plate. 

Why the difference? Well, in the former, that is, institution-based learning, barriers are set up *so that* opportunities and participation are reserved to only a selected few. You have to pass lots of screenings, interviews, admissions to gain access to resources and opportunities. In the open science regime that MLC operates in, there's no cost to entry, but you have to prove your worth through the *actual work of doing research*. Doesn't that sound like a fairer playground to grow, gain recognition and opportunities? If you are convinced, here is how each component in the research training looks like in MLC:

* find your own idea: e.g. browsing the pool of [RFPs](https://mlcollective.org/rfp/), attending [research jams](https://mlcollective.org/events/#jam) and [reading groups](https://mlcollective.org/dlct/), participating in open science events and initiatives, and actively thinking;
* seek out for your own collaborators: e.g. posting in `#seek-collaborator` channel on the MLC Discord, reach out to people that have published similar work, attend [research jams](https://mlcollective.org/events/#jam) to observe what others are doing and how they are doing it;
* start iterating on experiments, making plots. If you need computational support, email `compute@mlcollective.org`
* set up your own regimen with a lot of self discipline, e.g. set recurring meetings with your collaborators, commit to presenting at every [research jam](https://mlcollective.org/events/#jam) to push yourself to make decent progress;
* share your intermediate results in channels under the `RESEARCH IN PROGRESS` category, for example the `i-have-an-idea` and `i-made-a-plot` channels;
* start writing your first draft of the paper; ask in `write-submit-rebuttal` for help with feedback;
* set your own goal and follow through.

For each component there are existing resources in MLC, but you will have to be the one that *actively use them*. It may sound harder — indeed it is harder to have to self-motivate, self-examine, and take full responsibility of your path (instead of relying on an assigned advisor or manager) from the start, but we believe it is fairer, more transparent, and you'll come out a better trained researcher this way.