There are a number of places on the Web where you can see the current whatever, for example, the current World Population. A recent favorite is http://www.worldometers.info/.

The set up was:

• Show kids the site, let them see the number
• Tell them a convenient future time that’s during a future class (e.g., 9:15 AM on March 17), maybe two weeks away.

The task was:

• Predict the population at that time. You must turn in your prediction during the class before the one you’r predicting for.
• Say you’ll give a prize for the closest result.

What is cool about it:

It’s all about rates, but has all sorts of juicy properties such as many-digit numbers that, as far as they know, need to be precise to the nearest “one”; and of course the issue of time. It owuld be convenient if they took data at 9:15 AM for a few days, but that’s not practical. So they have to convert to some uniform unit: decimal days, whatever, to figure out what to multiply by.

What happened:

No one (duh) made a graph or set up the equation for a line. But everyone computed a rate, such as people per hour.

The best figure was within a few hundred, but answers (in this “regular” algebra class”) varied by millions.

Next step would have been to collect everyone’s data, use a good tool such as Fathom, plot it, and see if we could, as a class, make a better predction for the next class. Of course there was no time for that this time around.

They’re seriously gigantic, and erudite, and (like some recent work in Texas over the summer by Jennifer Smith, thanks, UTeach!) these people are actually solving the universe.

The thing is, I originally thought that with the money being a scarce resource, no community of scientists would ever imagine actually solving it! And I liked that in that it was “like science”—you can learn a lot about the problem, but you never fully solve it. Furthermore, I liked having to deal with the frustration of not knowing the answer…and having that be like science too.

And yet, and yet. The four-color universe has always been problematic that way: it smells like a math puzzle. People say it’s like two-dimensional Mastermind. So I have also claimed that the frustration of not getting the answer is perhaps enhanced because it seems like something that you solve: a puzzle.

When we first got the Critters scenario working, I had the impression (without really doing any research) that no one thought that they should “solve” it. It was normal that they would learn about the answer without finding it. It smelled different from the colors.

I asked David Brookes about this, asking what criteria he used to give them money (they needed a lot to do their solution!). He replied, in part.

…But generally I was fairly soft-hearted and if they contributed and asked, I gave them some more money.  Some people really got the observe-hypothesis-test sequence really well and you can see that in some of the papers.  Others think it is this giant puzzle they have to solve.  I didn’t think of making it so that they didn’t come up with a solution…next time I’ll do it better!

Of course there is no “should,” no “better!” If the students are engaged and observing, making hypotheses, and testing them, heck, that’s what we’re here for.

But I’m interested in this issue. What makes a compressed simulation experience like the ones we can support feel and act like science? And how much does it matter how authentically scientific it is?

I am gratified by those who have been enthusiastically using the prototype Nature of Science simulation software, and we will be moving into the long-promised new version eventually, really! But as these things sometimes go, many things have gotten in the way. And there they still are. So I’m not making projections right now just when things will move forward.

Fortunately, the old version is still working pretty much OK with a few burbles, most notably the old bug in the “Critters” scenario in which you get two new critters every time you ask for just one.

Anyway, thanks for all your patience!

—Tim

Sorry, but you now need to be logged in to comment. I have been spending far too long every day deleting all the ads that robots place in this blog. Don’t let that stop you, though!

That is, it’s only about the structure or grammar of the language, not its meaning. You might reasonably ask, what’s the point of simulating learning a language if you don’t know what the language means?

First let us say that it would be really cool if we could study meaning as well as structure, but for now, that’s too hard for this computer system. So what do we get out of syntax alone?

Quite a bit, as it turns out.

Think of language be a metaphor for something else. Chemistry, for example. The fundamental question in the simulation is: What patterns of words—what strings of colors, what sequences of parts of speech—are grammatical? In chemistry, one question is: What patterns of atoms make molecules?

Learning that any number of reds can precede a blue is analogous to learning that Carbons can bond with up to four neighboring atoms. And just as in chemistry, our linguistic model is based on an underlying explanatory structure. In chemistry, we might explain bonding based on an understanding of atomic structure and electron configurations, or we might explain it by saying that atoms are little balls with differing numbers of holes. Either one can explain why Carbon can bond with four other atoms, and can explain why some combinations of atoms make molecules and others do not. Discovering how many atoms each element can bond with has nothing to do with what the resulting molecules mean—what they are useful for, or what they react with—but this “chemical syntax” is an important thing to discover. And that’s what we’re simulating here.

There is, in fact, an underlying explanatory structure to the grammars in this scenario. I’m reluctant to let it out (although I will in a later post) because in science, we never know the real mechanism, the real explanation. We never know the Mind of God. What matters is how well our explanations do at prediction. We think atoms have electrons (as opposed to being balls with holes) because the balls-with-holes model does not to as good a job of explaining how real materials work as the one with electrons.

Does that mean electrons are real? Nope.

If it works, does it matter? Nope!

They’re not really words—they’re parts of speech. If you think of blue-red-green-blue as noun-verb-adjective-noun, it may make more sense.

Eventually, we’ll make an (optional) expansion of the scenario to include the words themselves. Suppose all worlds ending in vowels are red; all other words three letters long or shorter are blue; and among the rest, those beginning with “th” are green. Thus “pole” is red, “cat” is blue, and “polecat” is not a word—and your native would tell you. You would then have two tasks: learning what constitutes a word as well as what makes a sentence grammatical.

Why did I set up this elaborate computer-based system so that you have to enter data by hand? There are three basic answers.

• First, the more mundane: it’s a good thing to get practice reading scales. The scales for the thermometer and the temperature sensor are different, and force you to estimate the decimal parts of the values differently.
• Second, it’s good to have data that (like real data) are not perfect. One way to accomplish this in a simulation is to add a little randomness to the computer values, but another is to introduce a human element. If you read the temperature as 27.3 and I read it as 27.4, our graphs will be slightly different.
• Finally, and most important for learning, typing the data slows you down, and blocks a change-click-change-click cycle of data collection. In a computer simulation, we have a tendency to collect data as fast as we can and think later. And why not? It takes no effort (or thought) to collect it. In a real lab, however, it’s not only more costly, but more trouble to set up the equipment for another data point. So if it takes some effort to get that next point, you may be more prudent and more thoughtful in deciding what point to collect. Not that it’s all that hard to type another value!

Of course, sometimes you already know how to read scales, and you just want more data, faster—and make it perfect, please! I wish there were an easy way to customize the scenario, but there isn’t (yet, anyway). So whoever designs the scenario (in this case, me) decides, as a matter of taste, what data you have to type in and what data just appears.

Oh: and the gauges themselves? It’s hard to see the indicators that tell you where the number is. I know that; I’m hoping that a better Flash programmer than I will fix it. In the meantime, revel, as you squint, in how well the graphic simulates measurement error.

I have just posted a new build of the system at http://www.bigtimescience.com that fixes a few bugs. Most importantly, you can now get an email from our system when you sign up for a subscription account. Still doesn’t work with pay-as-you-go, but we’re working on it!

Also, although I am not supposed to be writing new scenarios, I have posted prototype code for a “linguistics” scenario that, in a way, is the most deeply scientific of the whole set.

Here’s the idea: you’re a linguist learning a new language with the help of a native speaker. You can listen to the speaker and hear a grammatical sentence, or you can speak a sentence and learn from the native whether what you said is grammatical.

Your goal is to learn how to distinguish grammatical from ungrammatical sentences. Of course, it’s not random: there are rules underlying the language. These grammar rules are analogous to the underlying rules of science, and telling what’s grammatical is analogous to making a prediction.

The current grammar is pretty complicated (I’ll randomate it eventually and make it so you can control how complicated the grammar is), but sentences are simple: they are a sequence of colors, red, green, and blue, just the initial (capital) letters. So a sentence might be RRBG for “red-red-blue-green.” Don’t use a period at the end!

Underlying the model here is a transformational system just like the one in the excellent game Queries ‘n’ Theories by Joan Ross and Layman Allen, on which this scenario is based.

Make a world and try it!

See the commercial game.

• To make new worlds, you need to register. We especially need a good email address so we can contact you.
• To be a “player,” you don’t need to register. In a classroom, for example, a teacher should make the world using the Class or Workshop scheme, and write down the (temporary) usernames, one for each team. The username is the same as the password. Hand one to each team.
• There is a new scheme for attaching data to papers. The basic idea is that instead of selecting the data piece by piece from a long list, you put the data into a graphic and then include the graphic as a figure in the paper. Details:
  1. Make a graph (or other visualization) of your data.
  2. Take a snapshot of the image.
  3. You will now get to see all your images in the Manage Images panel in View Data.
  4. When you compose a paper, press the little “image” button. You’ll see a palette of all your images.
  5. Choose the image you want and it will be inserted into your paper as a figure.
  6. All of the data used in the image will be attached to the paper.
  7. To import data from a paper you’re reading in the journal, press the link you will see there.
• Similarly, you do the same thing for citations, but press the other button. You will see a list of all the papers that you have read. Clicking it inserts a citation at the cursor and adds the paper to the list of references.
• Some functions require that you be a “premium user.” You will be asked for money. Don’t believe it. Just proceed, or make up a credit card number (I use “6″ or “37″), or claim that you will send a PO. You will be accepted. The system will email you a security code which you should paste into the box you will find. The deal is, you are agreeing during this beta period to give us feedback.

So enjoy! Let me know how it goes!

—Tim

Our system deliberately slows down the investigative process as well. And the chief way we do this is with money.

For example, the four-color universe is, deep down inside, a math puzzle, not a science setting at all. But you can’t treat it like a puzzle because you run out of money. This means that instead of figuring out the whole pattern in the universe, you have to be content with finding out things about the pattern. In addition, since you have such limited opportunities to do experiments, you had better be really careful about choosing what experiment you do.

That is, you have to slow down and think. You have to reflect.

Writing papers does this too, in a different way. If you want to get published, your paper has to be good enough, so (depending on your editor) you have to stop and put some thought into it.

As with all things, though, it’s a balance. The drive to solve the problem seems to push many students to try to be fast, to keep going. This is good as long as they reflect! Some scientists (often adults), however, take reflection to an extreme and have trouble getting out of the blocks.