[dandanisaur]

Moving to an area specifically for your startup, even if you were told to do so, can really show your dedication (not just by saying it). You are proving that you are willing to do whatever it takes. I've known many who have moved for the 3 months @ YC and then have rightfully moved back... I don't think any of them have regretted. YC may be looking for someone who can make it work, if there is a will, there is a way. THAT is what they are looking for.

There is a method to the madness. They have proven it.

[mozumder]

One problem is that, for a lot of industries, San Francisco is the wrong place to be.

If you want to be in fashion-tech, for example, you want to be in New York (or possibly London). And, there are lot of other industries concentrated in specific areas around the world.

Limiting to San Francisco is limiting your industry. Sure you can get a ton of web developers, but you'll lose access & relationships to the REAL industry insiders that matter for your vertical.

[mjn]

I think that can be true, but YC has a model that works for them, so it seems ok to me to not try to cover every industry in the world. Some of them are quite different, to the extent that YC's networks and expertise might not be as immediately useful anyway. For example Houston has a lot of energy-technology startups, especially in oil & gas, doing anything from the "hardware" side to analytics to GIS type stuff. But it's basically its own world that has little overlap with the Bay Area startup world. Different investors, different types of customers, different production challenges, different employee demographics, etc.

[onion2k]

YC has proven an accelerator works when teams move to be part of it. That doesn't prove that investing in companies that are based elsewhere and take part remotely doesn't work though, or that moving was a contributing factor to success, or that founders who are willing to move are demonstrating a greater level of dedication (just a different sort of dedication).

To know whether remote will work YC needs to try it.

(Mind you, even with YC's intake, it's likely you'd never see a statistically significant result either way. There's too many variables and not enough samples.)