Introducing Guesstimate, a Spreadsheet for Things That Aren’t Certain

I wasn't expecting this to go on hnews yet, but happy to take any questions!

I would be much more interested if this product were &quot;choose probability density function centric.&quot; Then, the Monte Carlo engine would gain much more interest. Being able to choose or specify arbitrary distributions, and then run simulations, would be valuable.<p>Of special interest are non-continuous distributions. How often have normal distribution reasoning failed in finance? Put another way, a user should be able to model a distribution himself.

Relevant—Uncertain&lt;T&gt;: A First-Order Type for Uncertain Data (Microsoft Research)<p><a href="http:&#x2F;&#x2F;research.microsoft.com&#x2F;pubs&#x2F;208236&#x2F;asplos077-bornholtA.pdf" rel="nofollow">http:&#x2F;&#x2F;research.microsoft.com&#x2F;pubs&#x2F;208236&#x2F;asplos077-bornholt...</a>

I&#x27;m convinced that an excel sort of lay-person&#x27;s computing platform is where probabilistic programming will really take off. This seems really cool!

It&#x27;s not really related but it made me think of a friend&#x27;s PhD thesis on uncertain data. If the subject interests you, be sure to checkout the summary of his (impressive) work: <a href="http:&#x2F;&#x2F;a3nm.net&#x2F;blog&#x2F;phd_summary.html" rel="nofollow">http:&#x2F;&#x2F;a3nm.net&#x2F;blog&#x2F;phd_summary.html</a>.

I like it. I had to do a strategy session with a client a couple of weeks ago and we needed to estimate how much the strategy was likely to cost over the next few months the. We had quite a few variables to work with though. This would have been handy in such a scenario I presume? We knew what are components and the ranges were.

This is very similar to the paper <a href="http:&#x2F;&#x2F;www.isi.edu&#x2F;~szekely&#x2F;contents&#x2F;papers&#x2F;2012&#x2F;szekely2012-iui.pdf" rel="nofollow">http:&#x2F;&#x2F;www.isi.edu&#x2F;~szekely&#x2F;contents&#x2F;papers&#x2F;2012&#x2F;szekely2012...</a><p>As per the paper , you can choose arbitrary distributions , construct a fluent graph , run Monte Carlo simulation and get the result - |via <a href="http:&#x2F;&#x2F;bit.ly&#x2F;hnbuzz01" rel="nofollow">http:&#x2F;&#x2F;bit.ly&#x2F;hnbuzz01</a> |

&#x27;Fuzzy logic&#x27; seems to be an ex-buzzphrase nowadays, but this seems pretty close to that territory. A variable&#x2F;cell&#x2F;logical-unit containing not a single value, but a distribution (often between bounds), and getting combined with other similar variables&#x2F;cells&#x2F;logical-units in ways that understand and respect the probability distributions.<p>Perhaps that field can provide a potential source of new names, when you decide to market this as a company.

Very much like Crystal Ball - an Excel add-on that&#x27;s popular in the finance and energy fields.

Direct link to Github: <a href="https:&#x2F;&#x2F;github.com&#x2F;getguesstimate&#x2F;guesstimate-app" rel="nofollow">https:&#x2F;&#x2F;github.com&#x2F;getguesstimate&#x2F;guesstimate-app</a>

I was watching &quot;Total time spent watching this video&quot; video, and had a basic question.<p>How does one tell guesstimate that there&#x27;s a hard lower bound on a quantity. ie. Video Length is at least 0, because negative watch times are unphysical? I know the specified distribution in this case is very narrow (the video lasting between -1 and 0 minutes has probability ~0.000032). But the answer does come out to be 26±32, which includes a substantial unphysical region.<p>And, if I give a hard lower bound on Video Length, can it propagate that knowledge into an asymmetric error on Total time?

Awesome! I use Crystal Ball (<a href="http:&#x2F;&#x2F;www.oracle.com&#x2F;us&#x2F;products&#x2F;applications&#x2F;crystalball&#x2F;overview&#x2F;index.html" rel="nofollow">http:&#x2F;&#x2F;www.oracle.com&#x2F;us&#x2F;products&#x2F;applications&#x2F;crystalball&#x2F;o...</a>) with triangular distributions and Monte Carlo for software project cost estimation. Crystal Ball costs thousands of dollars so I will be following this with interest.

I think this is super cool! We&#x27;re so bad at estimating probabilities (think Han Solo&#x27;s &quot;never tell me the odds&quot;) that this helps visualize the distribution of outcomes

This is really cool. Can anyone recommend any particularly good&#x2F;cogent Simple Caveman explanations of how Bayesoan theory&#x2F;Monte Carlo simulation work?

It seems that using montecarlo in excel is apparently pretty commonm, but I&#x27;m not sure if it can be done without addons.

A &quot;Rumsfeldian&quot; visualizer. Neat!<p><a href="http:&#x2F;&#x2F;www.theatlantic.com&#x2F;politics&#x2F;archive&#x2F;2014&#x2F;03&#x2F;rumsfelds-knowns-and-unknowns-the-intellectual-history-of-a-quip&#x2F;359719&#x2F;" rel="nofollow">http:&#x2F;&#x2F;www.theatlantic.com&#x2F;politics&#x2F;archive&#x2F;2014&#x2F;03&#x2F;rumsfeld...</a>

What fun - I did a monte carlo estimate a few years back when trying to determine what purchases price of house my girlfriend and I could afford. It depended on probable interest rate, how much my old house would sell for, etc. It&#x27;d be interesting to see how simply it could be modeled in this.

Isn&#x27;t how all software should be written? Expressions that represent a set of all possible values, effectively replacing the need for types.<p>Surely, such a platform would make building an app 100 times easier. Not that building apps is a good use of our resources.

Interesting idea - but you could certainly do this in any spreadsheet application with multiple cells to represent ranges etc.. I think the of estimating probabilities issue can be considered to be more of a practices issue than a tools issue.

This is a really great interface, and cool idea.<p>You might consider upping the run count, or maybe narrowing your bins for the visualization. Either way, it&#x27;s great to see more tools embracing probability and uncertainty like this.

hey nice tool on a side note what tool did u use to create the animated tutorials on your git page:<a href="https:&#x2F;&#x2F;github.com&#x2F;getguesstimate&#x2F;guesstimate-app" rel="nofollow">https:&#x2F;&#x2F;github.com&#x2F;getguesstimate&#x2F;guesstimate-app</a><p>image link <a href="https:&#x2F;&#x2F;camo.githubusercontent.com&#x2F;8fd97a97fa656a1eb92294f0fc436885b5d8dbb3&#x2F;687474703a2f2f672e7265636f726469742e636f2f6c636b496670416b69412e676966" rel="nofollow">https:&#x2F;&#x2F;camo.githubusercontent.com&#x2F;8fd97a97fa656a1eb92294f0f...</a>

It&#x27;d be nice to be able to show &#x2F; enter n-dimensional histograms too, so that one can get an idea of &#x2F; control the correlation between two outputs &#x2F; inputs.

This is awesome. I am aware of the math behind, but this really made it much simpler to use. And also to make complex models.

Cool idea but doesn&#x27;t render into anything useful on my S5....

How are correlations between variables accounted for?

Finally!

Nice, but I can&#x27;t help thinking of spreadsheets as something of a crutch.<p>Also check out:<p><a href="http:&#x2F;&#x2F;probcomp.csail.mit.edu&#x2F;bayesdb&#x2F;" rel="nofollow">http:&#x2F;&#x2F;probcomp.csail.mit.edu&#x2F;bayesdb&#x2F;</a><p><a href="https:&#x2F;&#x2F;github.com&#x2F;taschini&#x2F;pyinterval" rel="nofollow">https:&#x2F;&#x2F;github.com&#x2F;taschini&#x2F;pyinterval</a> <a href="http:&#x2F;&#x2F;mavrinac.com&#x2F;index.cgi?page=fuzzpy" rel="nofollow">http:&#x2F;&#x2F;mavrinac.com&#x2F;index.cgi?page=fuzzpy</a>