Welcome to today's #GlobalScienceShow as part of #FUTURES2020

From now until 6pm GMT we have a new researcher every 20 minutes ready to share their work with you!

What's going on? Check out the schedule below for info on all presenters!

First up at 12.20pm is @JessieW_Palaeo

"You don't see radio changing gradually and suddenly become radar. #Darwin's theory doesn't work for #technology, and we have to make a new observation."

- W Brian Arthur at SFI today

#evolution #innovation #invention

"Novel technologies are constructed from existing technologies. These offer themselves as components for the construction of further technologies."

- W Brian Arthur at SFI today

#evolution #innovation #invention #technology #engineering #design

Unlike the two-parent #inheritance typical to (but not ubiquitous among) #complex organisms, #technology inherits features from n parents - a "vast ancestral #network" more like horizontal gene transfer networks in #bacteria.

W Brian Arthur on sumulating #invention on a chip:

Does #Archaeology support the Theory of the #AdjacentPossible - that #innovation is combinatorial, persistent, cumulative?

In short, yes. SFI's Tim Kohler on the #Paleolithic data of #technology #evolution & the #prehistory of the infamous "hockeystick" curve:

😉 @SAPIENS_org

#Economist Roger Koppl of @SyracuseU presents a model for the #AdjacentPossible at SFI.

Output = "a combinatorial explosion" or #singularity, albeit one that arguably occurred millions of years ago, or in the distant future (since the model doesn't specify t = 0):

Here is another very enjoyable conversation, with @Pandata19’s scientific advisory board member, Dr Jay Bhattacharya. Key ideas in this thread.

We are making world-changing decisions on the basis of evidence that is not very good. Vast scientific evidence tells us that infection fatality rates are much lower than originally expected. A small fraction of people get severe illness. 2/10

The scientific community has been resistant to evidence not supporting the majoritarian view, preferring instead to gin up panic, focusing on the worst case for everything the virus does & the best case for everything lockdowns do, and ignoring the range of uncertainty. 3/10

Academe is a strange place now, with debate stifled. But there's a sense some are opening up to considering opposing views. This is key, since suppression of views stops knowledge from progressing—the end of science. Public health norms of unified messaging complicate this. 4/10

Why did we, from the start, assume we knew nothing about this virus, instead of assuming a reasonable prior? Low susceptibility was evident early on. Assuming any virus is new is hard to square with our deep-time co-evolution with viruses, & their slow evolution. 5/10

1/

Get a cup of coffee.

In this thread, I'll help you understand the basics of Binomial Thinking.

The future is always uncertain. There are many different ways it can unfold -- some more likely than others. Binomial thinking helps us embrace this view.

2/

The S&P 500 index is at ~3768 today.

Suppose we want to predict where it will be 10 years from now.

Historically, we know that this index has returned ~10% per year.

If we simply extrapolate this, we get an estimate of ~9773 for the index 10 years from now:

3/

What we just did is called a "point estimate" -- a prediction about the future that's a single number (9773).

But of course, we know the future is uncertain. It's impossible to predict it so precisely.

So, there's a sense of *false precision* in point estimates like this.

4/

To emphasize the uncertainty inherent in such predictions, a better approach is to predict a *range* of values rather than a single number.

For example, we may say the index will return somewhere in the *range* of 8% to 12% (instead of a fixed 10%) per year.

5/

10 years from now, this implies an index value in the *range* [8135, 11703]:

Next up at #enigma2021, Sanghyun Hong will be speaking about "A SOUND MIND IN A VULNERABLE BODY: PRACTICAL HARDWARE ATTACKS ON DEEP LEARNING"

(Hint: speaker is on the

In recent years ML models have worked from research labs to production, which makes ML security important. Adversarial ML research studies how to mess with ML

For example by messing with the training data (c.f. Tay which became super-racist super-fast) or by foiling ML models by changing inputs in ways humans can't see.

Prior work considers ML models in a standalone, mathematical way
* looks at the robustness in an isolated manner
* doesn't look at the whole ecosystem and how the model is used -- ML models are running in real hardware with real software which has real vulns!

This talk focuses on hardware-level vulnerabilities. This is particularly interesting because these can break cryptographic guarantees (because those are outside of their threat models)
e.g. fault injection attacks, side-channel attacks

In physics, energy is that which is causally invariant or symmetric with respect to time. Said differently, energy is never created or destroyed, it is conserved in the universe.

Now there is a question about information. Is information causally invariant? Does information remain the same with respect to time? To understand this, let's first understand what energy is, then attempt to understand what information is.

The strange thing about energy is that it is this universal currency that seems to be interchangeable between the different forces in nature (i.e. electromagnetic, strong, weak and gravity). It is that thing that makes change possible.

Now let's assume that the universe is computational in nature. That is, anything we see in nature is a consequence of computation. That is, there is computation underneath the physics that is driving the universe to change.

What does it mean then that information is conserved in this computational universe? Also, how is energy conservation related to information conservation?