Cofounder Here: Happy to see this on hnews again.<p>Update: Matthew (the other cofounder) and I got Guesstimate to a stage we were happy with. After a good amount of work it seemed like several customers were pretty happy with it, but there weren't many obvious ways of making a ton more money on it, and we ran out of many of the most requested/obvious improvements. We're keeping it running, but it's not getting much more active development at this time.<p>Note that it's all open source, so if you want to host it in-house you're encouraged to do so. I'm also happy to answer questions about it or help with specific modeling concerns.<p>Right now I'm working at the Future of Humanity Insitute on some other tools I think will compliment Guesstimate well. There was a certain point where it seemed like many of the next main features would make more sense in separate apps. Hopefully, I'll be able to announce one of these soon.
Maybe it's just the short video and the FAQ, but I found it particularly difficult to find information about the distributions involved and how to choose that.<p>I imagine there a bunch of cases where the defaults would not work like you're trying to do error propagation (all normal distributions) or you're trying to compute interval arithmetic.<p>Is it the case that if you input a range which span multiple orders of magnitude then you get lognormal rather than normal?<p>I might not be exactly the target audience, but I would appreciate a more in-depth of the math and heuristics involved<p>EDIT: I found this on their blog<p><a href="https://medium.com/guesstimate-blog/lognormal-normal-833bf413c7a3" rel="nofollow">https://medium.com/guesstimate-blog/lognormal-normal-833bf41...</a>
I saw this a couple of years ago, when it was just a project. Now that there's a price, how did you guys decide on a price? How did you find your first customers? For a broadly applicable tool, how did you know where to start looking?
Does this permit Bayesian inference? e.g. looks like graphical probabilistic programming (hooking up various distributions and performing inference), except the key missing component is the ability to observe values for any given distribution beyond the prior.
I'm developing a similar open-source app for statistical modeling and inference in the browser: <a href="https://statsim.com" rel="nofollow">https://statsim.com</a>. You can create probabilistic models and then infer their parameters using algorithms such as MCMC or Hamiltonian Monte Carlo. The app is still in beta but it might be useful. Some models: <a href="https://github.com/statsim/models" rel="nofollow">https://github.com/statsim/models</a>
This looks fantastic.<p>I love that it was a no BS signup and start using. Super clean and easy. It would be great to be able to show data on GIS as well - effectively showing the outcomes geographic representations. Ill see if the data I was looking to work with today will work with this tool meaningfully.
I've used Palisade @Risk quite a bit, but for my use case, most of the time I feel like I'm taking a Lamborghini to the comer store. This is perfect for someone like me who is more of a "casual" estimator of things modelling with probability.
God dammit. This sort of thing pisses me off. Here I am, on vacation, waiting for my family to wake up. What better way to spend my time but pursue HN. I happen upon something like this. Something so damned useful that I have no choice but investigate.
I would love to see this idea translated into event planning/calendaring. Probabilistic party planning. I want to see what might be happening tonight in addition to what is definitely happening.<p>"If 5 people show up at my house tomorrow evening, I'll hold a poker night." 10 people were invited and 4 of them RSVP yes and 2 of them RSVP no. It looks like there's a 95% chance I'm holding a poker night tomorrow.<p>"The X team has a monthly meeting on the 1st, never fail. They haven't decided on the location yet, just that it's on the North Side." As the team members pick possible locations, the possible locations appear more distinct until one is chosen.
It wasn’t obvious from the landing page but can you link estimates from different models? It would be super cool to directly import variables and their estimates from other models.
I proposed writing something like this while working for DuPont's Encirca platform. Years later still little to no adoption in the farm IT field of these models.
From 2015: <a href="https://news.ycombinator.com/item?id=10816563" rel="nofollow">https://news.ycombinator.com/item?id=10816563</a>