985 karma · joined September 15, 2010
It is harder to have the same statistical confidence of efficacy and safety in such studies, but clinical researchers try to address those issues by varying e.g. dosage quantities, time in between doses, etc.
Source: place I work is currently doing studies of this nature, and in general such studies seem to be well-understood and accepted by the FDA.
Imagine a hypothetical universe where, in order to watch one of your movies, people had to a) pay you $1 and also b) let you punch them in the nose. Then, when people sensibly start pirating your content instead because they don't want to get punched, you loudly proclaim that they are stealing the $1 you are owed.
That's what's happening here. People want to watch your content, and are willing to pay for it. But they don't want to pay for it AND get punched in the nose. They pirate because your distributors, and by extension you yourself, have made it impossible to watch your content (and pay you!) in any other way.
I get that your natural rejoinder will be "if the content is not worth being punched in the nose, just don't watch it!" Which is fair. Debatable, but fair. Just don't come here pretending that all you have asked for is the reasonable sum of $1 when you are actually demanding that your customers subject themselves to the indignity of your fist.
> I had a hard time thinking strategically about product. I was too busy supporting the entire organization on all things related to the product.
so that's a pretty clear indication of what was missing. Instead of being able to recognize broad industry trends and come up with broad product evolutions that meet those needs, this person is stuck answering quibbles about particular customers or particular implementations.
So, serious question: why do you have a LinkedIn account? I know there is theoretical benefit in maintaining connections to professional acquaintances, but in sticking to email and a few alumni Slack channels, I have never felt limited in that department. What is the benefit of your LinkedIn account and why do you keep it?
I could go on and on about the problems:
1) I (and the two other programmers I was working with) had a job. The pace at which we rolled out features and bugfixes was glacial, even when customers were clamoring for them. And this statement is just about improving the app; it doesn't even consider the other tasks involved in running a business, on which I was even less excited to spend effort.
2) Pharmacy management software is a fragmented market, and almost every player is impossible to integrate with. Since our product didn't do the insurance, billing, inventory management, etc. it meant we had to integrate with them. We started out by telling people to run them side by side and manually make changes to both our app and that of their vendor. Not a great pitch. Looking back, I wish I'd invested in this from the get-go, instead of trying to do it after building the app.
3) The product required pharmacies to change their entire workflow to even realize the benefits. Which is fine, but we sold it as a software thing that worked magically, and didn't invest at all in helping customers use it successfully.
4) Our cofounders (pharmacists) were not interested in running a SaaS company. They viewed the app as their strategic advantage in their own pharmacy, and dragged their feet on the customer outreach we hoped they would take ownership of. I felt like they were actively working against me. This was the biggest issue, and what caused me to finally decide to close it down. Someone else in this thread made a statement of "I went from the prototype guy to the only one who cared." and that statement really resonates with me. In retrospect, I had the opportunity to structure the deal as more of a product/customer one than a cofounder one, and I wish I had been smarter about that.
I still think it was a great idea, and we built a slick app. I just failed on the sales and customer outreach side of the business. The app still exists in a zombie state, and a few customers still use and pay for it. I haven't done anything on it other than renew SSL certs since 2019.
I didn't read the report by Hired that this article was based on (because it's useless blogspam) but I did search for the word "control" to see if they were controlling for these sorts of confounding variables, and I came up empty-handed.
This is a common canard. I can see how it might make some intuitive sense, but it appears to be utterly mistaken. When I brought up the idea at my current workplace, we ended up choosing not to pay candidates, because almost 100% of our female software engineers declared that they would not have started an interview process where they were being paid unless they were VERY certain they were qualified for the role.
The fourth scenario is more interesting because we've admitted the distribution of p is unknown to us. You note that if p is uniform, all intelligent lifeforms should conclude they live in a universe where p is high, but I'm not certain of that. It seems to me that this logic only works if you conclude there is _one_ universe for each possible value of p. But if, for each possible value of p, the multiverse contains universes for each possible outcome of a Bernoulli trial with probability p, then we can't conclude anything at all. To see this, notice that the number of Lucky universes (where p is high) is uncountably infinite, just as the number of Unlucky-Lucky universes (where p is low but a miracle has occurred) is uncountably infinite.
I haven't thought this next part through as deeply, but I believe this problem compounds given the unknown distribution of p. This is because subjectivist models of probability (e.g. Bayes' Theorem) are heavily influenced by the distribution of the prior when we have only observed one trial. And in cases like this one, where the only trials that _could_ be observed are successes, I believe it's all but entirely driven by the distribution of the prior. (You could imagine a hyperprior - a distribution of distributions of p, but I don't think we have any good reason for deciding what that would be either).
Anyway, thanks for this. I love geeking out about epistemology.
> If X occurred, and is more likely under hypothesis B than A, then X is evidence for hypothesis B, otherwise you don't have a consistent system of probability.
The point is, this idea breaks down when neither hypothesis would exist if X had not occurred. This differs from other questions of probability because the conditional probability of an event is predicated on the ability of an observer to ask about it. We should indeed be indifferent between the two hypotheses, if our own existence is all we have to go on. (This might seem strange, if you think of hypothesis B as having 100x the prior probability, but the point here is that we really don't have a basis for a prior at all.)
> Now you can further imagine that God comes down from heaven and tells you that in fact both of these two universes actually exist, and they are the only two universes that exist, so you are definitely in one of the two, but doesn't tell you which one. Is your posterior probability now 50% between the two?
No, of course I would assume I was in universe B. This is a very different question. Instead of "what is the nature of our universe" the question is "of the two existing universes God has described, which do I occupy?"
1) There exist billions and billions of stars, each of which can be thought of an independent Bernoulli trial with probability P as to whether intelligent life exists in that star system.
2) P must be a non-negligible quantity, because we ourselves exist
3) Therefore intelligent life "must" exist elsewhere in the universe due to the sheer number of trials.
This is an epistemological error that echoes the sort of error made by proponents of intelligent design. They conclude that the chance of thinking beings evolving is prodigiously unlikely, and therefore our existence is proof of a creator.
But in fact, you cannot post-hoc conclude that our existence says anything at all about P, because if we didn't exist, there would be no one to ask the question. Therefore the fact that we are wondering about the value of P means our existence is an implicit axiom of the logical system itself. One way to think of it is there could be infinitely many possible universes with no life at all, and by definition the question of P can only discussed in the one branch, in all that lifeless space of possibilities, where thinking beings actually exist. So P could, in fact, be made arbitrarily small, and Earth could indeed be the only place in the universe with intelligent life.
Note that this doesn't claim that extraterrestrial life _doesn't_ exist. It merely points out the truth in all of this P-hacking: that we have no evidentiary basis for believing one way or the other.
When you double count like this, you are implicitly assuming deforestation ALLOWED the polluting business to occupy the new land, and they wouldn't have polluted to the same degree elsewhere (because it wasn't available, wasn't as profitable, or something else).
A totally reasonable claim! I could explore it further. But IMO one should make this claim in good faith, make it clear what you're proposing, not engage in this kind of statistical sleight of hand.
I am totally opposed to deforesting the Amazon. But I'm sick of political participants that think it's OK to lie to people, simply because the implications of the lie favor some greater truth.
When you ascribe the effects to the Amazon, you combine the two. As a simple exercise, imagine that the deforested land had been left fallow, and the CO2 emitting business had set up shop at a different plot of land. You wouldn't ascribe the CO2 emitted by the business to the Amazon, and therefore, when you DO ascribe it to the Amazon, you are double-counting the effect.
None of this is to say that it's okay to deforest the Amazon. I just hate it when people lie with numbers for the sake of a catchy headline.
They're double counting the deforestation. There's nothing else to see here.
We've been doing a lot of remote work during covid and I have definitely noticed the decline of water-cooler solidarity. We still do hire remote engineers and data scientists, but only in cases where they are the literal top of their field.
[0] Standard disclaimers: I work at Recursion, have drunk the Kool-Aid, etc etc
If you're exploring the world of AI for drug discovery, IMO, it's important not to fall for the trap of companies that just do "AI on public available data". That's just the same "the fundamental problem of software is a lack of ideas" phenomenon the article was talking about. You want to meld software and data science skills with the fundamental work of whatever industry you want to be the vanguard of.
Seriously, if you're in the market, check out Recursion. My contact is in my bio also, and I'll answer any questions Imran can't (or won't :p).