Tell HN: Here's how my team got to the YC interview stage

We've fought pretty hard over design details—we're all very passionate about design and perfectionistic when it comes to the user experience, and sometimes our aesthetics conflict. I don't anticipate any more serious fights, but I've read enough startup stories to know that such things are always possible. I'm not sure that what you're suggesting is right, though. Are you saying that a startup that's been torn asunder by internal conflict and reunited by the founders' convictions is stronger than one where work has proceeded smoothly and uneventfully?

And yes, I do have that drive to push through. It's not every day that you get the chance to change the world. I'm not throwing that away, no matter how difficult things might seem.

As to "soft and nice"—well, thanks! I get what you're saying, but my impression is that knowledge >> toughness in the startup world. Being tough is mainly about understanding that you have other options, that you don't have to accept everything that's put in front of your nose.

I agree - you seem like a smart cat, best of luck with YC.

That would be absolutely awesome. Of course, in principle, there's nothing preventing researchers from doing that now. I think that being able to actually manipulate things on the same website that the data exists on will be key to bringing "open source papers" into the academic mainstream.

Perhaps this could work well with results based on publicly available data sets (like the Protein Data Bank at www.rcsb.org), but I'd be surprised if this isn't done already to a significant degree. Labs post programs (or their biological equivalents, protocols) on their websites all the time.

Don't forget that data can be polished as well - for a hypothetical cheater, you'd simply be moving the "messaging" work from the finished project to the source files you mention above. There's a lot more to trusting a paper then just trusting the figures; you're trusting the scientific process that a lab is undertaking, and it's almost always impractical to require proof of every stage of that process.

Another downside is that not every scientist is ok with giving access to all of their hard-earned data - that opens up analyses and potential scoops to other groups. This is especially true in biomedical/clinical studies, where data is very expensive to gather and can take years to analyze. Not saying raw data is never shared, but there is a balance there.

Finally, scientists greatly prefers testing a published claim by pursuing a follow-up hypothesis rather then redoing an existing analyses. That way you kill two birds with one stone. It's inevitable that some irreproducible results slip in with this system, but if the claims are significant enough sooner or later someone who's trying to build on the claims (and failing) will go looking for more details and the fraud is exposed. The fear of being exposed in that way is what keeps the system honest, and to a large extent this system works.

That's a good question, and figures are hard to find. Statistical analysis is a huge market; SPSS, for instance, was recently bought by IBM for $1.2 billion, and SAS made $2.3 billion in 2009 alone. But while those companies started out as humble makers of academic software, and continue to make significant revenues that way, they both found the business analytics market to be far more lucrative, so those figures largely reflect their sales of large enterprise solutions. Companies that primarily sell to academics (e.g. Stata, Minitab) are private and don't publish sales figures.

We think our most promising immediate markets are social scientists (of which there are about 25,000 in the United States) and stats/science classrooms. While the original concept was mainly aimed at researchers, we've been shifting more toward classrooms recently, based on the feedback we've been getting. The two products we envision—the research product and the educational one—share the same core, as far as the interface and computational backend. So, we believe that we can pursue both markets, much as Lingt (YC S09) has both an individual product and a classroom product.

I should emphasize that my blog post is in no way intended as a pitch for Theoryville reflecting our most up-to-date thinking about our product or our business model.