Hacker News Reader

Comment by Tangokat | Hacker News Reader

"And, in essence, this kind of sums up Watson for me. You need to tinker with it a bit before you can find something usable."

I think this is the most interesting part. Watson seems to work WITH you rather than for you.

Presently it appears to work very well as an inspiration and suggests things that are so far out you would never come up with them yourself. An example is the radish, cod, ginger, olives, tomato and green onions pizza.

I would've loved a bit more focus on what other people thought though. I got the impression that the writer was a bit biased to love the dishes before he tried them.

View on HN

"Watson seems to work WITH you rather than for you."

Absolutely right, and that interaction is the most interesting part. This also reminds me a talk about Palantir at GovCon. The speaker tells us two stories.

The first story was about Advanced Chess. We know that human and computer both are good at chess for two very different reasons. We also know they these skills can complement each other. So in Advanced Chess, instead of playing against each other, human and computer form a single team and fight against another team.

It's easy to think that if we take the single strongest chess player and the single most powerful computer, they might dominate the tournament. But that didn't happen. The team that won the tournament was a modestly skilled chess player with a regular laptop running freely available open source chess tools. So how did this happen?

It turned out there was a third factor nobody consider. It was the quality of interface between the man and the machine. How easy it is for a human to develop the strategic oversight, the guidance and to rely on the computer for tactical component. Focusing on, on what they are individually good at.

The second story is about Paypal. At the time of Paypal, several other companies were trying to get involve in online payment industry. The other companies were CitiBank, eBay, Western Union. But one by one they folded back where the Paypal succeeded. What happened?

It turns out large corporations were trying to develop a better system to automatically verify transactions. But Paypal went the other way around. They tried to increase the effectiveness of their human analyst by providing better computing service. Everything on the computing side was to empower/help the analyst to explore his data faster, detect anomalies and to make a quick/better decision. It's the man not the machine that had the final decision in Paypal.

I think the core idea of these two stories is how differently can we use our computing power, and how we are doing so far. I think most solutions are trying to develop a set of fixed answers, and hoping their answer are better then their competitors. But I think, it should be the other way around.

Computers are here to assist us. The design of a system should not be reaching a fixed the goal, instead it should help us; acting as a guide providing necessary information helping to make an informed decision. And if used properly, human can already attain superpower.

And I am really curious to see how IBM Watson fits into this bill!

absolutely, would be fantastic to have watson incorporated into sites like supercook.com - really mindblowing use of computational power for the benefit of humanity.

You mentioned Palantir's talk at GovCon. Do you know if it's available online?

Yes it is available on YouTube. Here : https://www.youtube.com/watch?v=f86VKjFSMJE

He tells us the stories within the first 10 minutes. In the following 10 minutes, he talks about what how Palantir's software fit into the picture.

And in the last 20 minutes, he demonstrates the software as an intelligence officer, who is tracking some bad guy across multiple countries. Information have been fed to the software by informants. And now he job is to organize those information into patterns, and trying to come up with a hypothesis about what these bad guys are doing, and what they will do in the future. Finally he shared his hypothesis with his colleagues.

Just like Advance Chess, they are trying to reduce friction between data and the data analyst.

> An example is the radish, cod, ginger, olives, tomato and green onions pizza.

actually this sounds similar to a lot of seafood crudo dishes i've eaten in many major cities.

olive, tomato and onions is basically a bruschetta topping or a basic tapenade/salsa type mixture (to go with the bread or the 'pizza crust'). radish goes well with protein. think horseradish on steaks, sushi, etc.

i think that dish would taste pretty good, if you ate it like bruschetta. or you could eat the toppings with a fork and dip the crust in olive oil. or just fold it up like a taco and dip the whole thing in olive oil.

i think the primary advantage of watson is it isn't constrained by pre-conceived cultural combinations that humans tend to stick to. also why 'fusion' can taste amazing, or be really horrible.

That 'pizza' sounds like something closer to http://en.wikipedia.org/wiki/Tarte_flambée

The only thing 'Pizza' is the bread/base, the rest is something else?

Radish, ginger, olives, tomato and green onions are quite general. They could go together with a lot of stuff, especially fish. In the Philippines one of my favorite restaurants serves a soup with a similar recipe. No radishes or olives and with different varieties of fish, clams and squid.

Reply on news.ycombinator.com