Evidence is at the heart of science. When we talk about what distinguishes science from other ways of knowing, we talk about evidence-based reasoning. Which means that we draw inference and conclusion and make theories about the world based on evidence. So what do we mean by evidence? What is evidence to a scientist? Scientists use evidence based on data. That data could come from the lab or a telescope, but it's based on observations that can be framed quantitatively, that can be described numerically, and they can be shared. They also have to be reproducible, it's not evidence if you were the only one who can obtain it and no one can ever check or corroborate what you have. Equally, it's not subjective. Evidence is not open to the eye of the beholder or an interpretation that depends on one person and the next. Interpreting the meaning of patterns in data can indeed vary from one scientist to another, but the data themselves are not subjective. Finally, the evidence is of course never perfect. That's because as we've described, the observations themselves are limited in number or in quality and we never have quite as much evidence as we want. That doesn't mean however that we cannot draw reliable conclusions in science based on the evidence we have. There is no science without evidence. It's not possible to make an assertion about the natural world without evidence to back it up, that's what we mean by evidence-based reasoning. Otherwise, it's just your opinion and someone else could have a different opinion. Science is a communal activity. So scientists must be able to share data and make their data openly available. This is a big issue in modern science because increasingly large amounts of data are sitting in private domains or behind firewalls, or not accessible to other scientists. It is hard for scientists to come to a consensus especially on an important societal issue without the data being shared. Science is driven by data and given that, there are three tiers of progress in science. The first is gathering more data for a particular scientific experiment. If I'm observing for example a faint galaxy with a large telescope, there are a limited number of photons I can gather in a reasonable amount of time, which limits what I can say about the galaxy's size, shape or color. So I want more and more data. I will ask for more nights on that telescope or seek a larger telescope to get more data to improve my knowledge of the galaxy, and so on in different areas of science. The second level of improvement or progress involves repeating the experiment. To be more sure of my outcome for this particular faint galaxy, I might go to a different telescope or use a different detector. If this galaxy was at the edge of detection, I'd want to confirm it probably with another observation. At the very least with the same telescope and detector a year later, but maybe even better with a different telescope or a bigger telescope. The third level of progress involves corroboration by another scientist, and in the end, almost all scientific progress relies on this level of corroboration, where their experiment is repeated by someone else using an entirely different apparatus or equipment. A historical example can help us see the limitations of data and how they play into our sense of progress in science. This series of graphics shows drawings made of Saturn by different people in the first 50 years after the telescope was invented, from the 17th century. Each of these little drawings was made by different scientist or observer using a different telescope. Telescopes were quite poor quality in these days, the time of Galileo. Optical quality was poor and the lenses were small and had imperfections. We can see Saturn looks quite different in each of these graphics. In some, it doesn't appear to have rings at all or moons, in others it appears to have handles, in some it appears to have rings of different orientations. They're so different. How can we look at these images, these observations and say which is correct and which is wrong? The point is, at the state of the art, at the cutting edge of technology and observation, we cannot be sure. It's only through improvements in the observations that we become sure what Saturn actually looks like. Of course flash forward now to the early 21st century, we have spacecraft in orbit of Saturn, the Cassini, and Cassini takes exclusive pictures of Saturn and its rings down to the finest detail. But at the time when the telescope had just been invented, we learned of Saturn rings for the first time and we were unsure of their details. This is how science works to improve knowledge. But in the early stages, the knowledge is contingent and the inferences are unsure. Murray Gell-Mann famous Nobel Prize winning physicist once said, "Research is what I do when I don't know what I'm doing." Now, that might sound like a joke or he's not being serious. But what he's talking about, is a profound aspect of science, that science exist at the boundary between what we know and what we don't know where uncertainty is necessarily high. That's the excitement of science too you don't know what you're going to find. If you were spending all your time or we as scientists spent all our time measuring things we already knew to better and better precision, we'd never learn anything new. We'd never make profound progress. So scientists often operate at the limitations of their data and their observations, where uncertainty is particularly high. Here's another example from astronomy. In a sequence of pictures, we see a simple measurement of the two angles on the sky that define the position of any star measured with a telescope. We can measure those two angles and put them on a graph or a piece of paper. It's the position of the star. But what happens if we measure the position of a star three times or 10 times or a 100 times, going off to a different patch of sky each time before coming back to the star? We of course find out that the measurement is not exactly the same. The dots or the measurements don't perfectly overlie each other. That's because there's uncertainty in our measurement of the position of the star. Some of that uncertainty comes from the telescope itself, it's drives, its motors, its alignment. Some of it comes from the sky and the atmosphere blurring the image of the light and refracting it as it passes through the Earth's atmosphere. So all of these uncertainties combine to produce what we see as a scatter plot when multiple observations are made. But if we look at anyone observation, how do we know which is the right observation? Where is the star really? It seems like a simple question to ask. If we'd only made that one measurement, the first one, all we could say is the star is here. Having more information, more measurements in this case, ironically makes us less secure, less certain. But it gives us a better understanding of the errors involved in the observation. The scatterplot produced in the graph with many observations, clearly has a central locus or a mean position. The mean of the X and the Y is the best measurement of the position of the star, Our best estimate based on multiple measurements. The scatter or dispersion of those dots represents the error or the uncertainty. This is at the core of how science works. In theories developed by Carl Friedrich Gauss in the 17th century, we assume if there's nothing but random uncertainty associated with our errors in the apparatus, that this is defined by what's called a Gaussian distribution. You may know it as the bell curve. In many observations in science, this is what goes on. But notice that we only were able to measure the best position and its uncertainty by having multiple observations, which told us something about the uncertainty in the measurement itself. This is essentially how all science from high energy physics to astrophysics works. In astronomy in terms of evidence, we have three completely distinct regimes. When we're talking about physical evidence or direct evidence, is really quite local. Remember, we have only stood on one other body in space and that's the moon, a quarter of a million miles away, and with the Apollo missions, we brought back a few 100 kilograms of rocks that are now sitting in the Kennedy Space Center. So we can inspect the moon material in the same way we could a rock on the Earth's surface, and be very sure about the conclusions we draw from it. This is direct evidence, but it only applies for this one astronomical object. We have visited with our spacecraft, a slightly larger set of objects including moons of the outer solar system. We've landed for example on Titan and we've recognized with a comet and an asteroid. This is fairly direct information. But except in the case of a comment, we haven't brought back samples from these objects. So our information is getting less direct. Sometimes we get lucky and we get free samples from further out in the universe. Meteors represent pristine solar system material that falls to Earth, and so gives us information on situations that actually maybe billions of miles away when the material formed. These are our free samples from the outer solar system. In all of these three levels of information or direct samples, we're talking about a maximum of about a billion miles. What we need to remember is that, in the universe, that's a tiny region. We live in a universe 46 billion light years in any direction away from the Earth in size, and for 99.999 and many 9s after percent of that volume of space, our information is indirect and comes in the form of electromagnetic radiation. So in astronomy in general, we're usually dealing with remote sensing not direct samples, and we're dealing with inference based on electromagnetic radiation. That's true of most of the regimes we'll be visiting in this course. Science is based on evidence. In astronomy, that means data or observations made with a telescope. Progress involves evidence-based reasoning. In other words, we do not make assertions in science without evidence to back them up. What is the nature of this evidence? This evidence has to be reproducible. It has to be reproducible by another scientist who wants to test your experimental or repeat your observation, and that in fact is the gold standard in science, to repeat the observation. Since all the data and the evidence has limitations, we want more data and that's why science is a data hungry enterprise. We always need better and better and more and more evidence.