Hacker News Reader

Comment by techdragon | Hacker News Reader

Its arguably a possibility that we are so unsure of the effect because we have scientists "cautiously creeping" so as to not hurt their future reputation.

There was anomalous scaling in the observed effects in one Chinese research teams results, but it has not been repeated because full reproduction and pushing past their power levels would cost some decent money.

Seems stupid the (possibly) most groundbreaking advance in propulsion ever made, can't get a million or two when we collectively gamble more than that each day at casinos lotteries and horses races.

We could quickly eliminate the question of "is this anomalous results from bad experiment design" with a bigger experiment that would make such flaws more evident. Yet we waste time instead of money.

View on HN

I think you've nailed it there. A professional scientist can't take this seriously because the reputational risk is so high.

Science needs to embrace failure more!

Nobody is allowed to fail or make mistakes any more, not just scientists.

Your past used to fade with time. You used to be able to move town, escape your failure, try again.

Now, any failure is forever emblazoned in shrieking tones online, and the only escape is a new name, and starting over.

It's not just science, it's everything.

>Seems stupid the (possibly) most groundbreaking advance in propulsion ever made, can't get a million or two when we collectively gamble more than that each day at casinos lotteries and horses races.

Perhaps they need to set up a crowdfunding campaign to get those people betting on their experiment, gambling on a science experiment rather than a game of chance.

Exactly, in order to have a long career in science, one needs to toe the line, and keep in the mainstream.

Somebody built this complicated device without already having any understanding of how it supposedly works, and you think it might work anyway? Talk about winning the lottery.

What's stupid is betting on stupid.

> What's stupid is betting on stupid.

What's stupid is making investments which are known to have an expected return less than 1:1. Playing the lottery isn't stupid because we don't understand how it works; playing the lottery is stupid because we do understand how it works, we know what the expected return is, and we know that it's a worse investment than just stuffing cash under a matress. This is entirely uncontroversial, and lotteries are run by profit-making entities (either private firms or as a "tax on the ignorant" by governments) whose entire viability relies on this fact.

In contrast, things like the EmDrive are high-risk high-return investments. Their expected return is more difficult to estimate than a lottery, since we'd need to estimate the probability of it working and the expected return if it did work. However, whilst an idea like the EmDrive may be controversial, the idea that spending a small proportion of investment on ideas like the EmDrive isn't controversial. There may be arguments over how much counts as "small", which ideas should be prioritised, etc. but this just goes back to the uncertainty of estimating the expected return. It's entirely uncontroversial to say that, if it works, the EmDrive would be an incredibly lucrative technology; it's also uncontroversial to say that it's unlikely to work. The tricky part is working out which term dominates the expectation: does the big thing (the potential return) multiplied by the small thing (the probability) result in a big thing or a small thing (the expected return)?

This isn't high risk though. It's low risk. However, just because something's low risk doesn't mean it's worth the investment.

The problem here is that you could make this same argument about almost anything, a la Pascal's Wager.

> This isn't high risk though. It's low risk.

I meant "high risk" in the sense that the return has a high variance, with all of it being concentrated in a thin sliver of the probability; apologies if I've misused a technical term.

> The problem here is that you could make this same argument about almost anything, a la Pascal's Wager.

I'd call it a consequence of expected return being a widely-applicable calculation, rather than a "problem" per se. Even if we knew the expected returns, we'd still need a decision procedure to perform the allocation.

My point is that it's uncontroversial to avoid putting all funding into the most promising project (e.g. other Physics research doesn't wait until we're "finished" with the LHC), so there's certainly scope for allocating a small budget to more "fringe" research like the EmDrive. I'm not in charge of research budgets, but as a simplified argument we might imagine allocating funding on an exponential scale, based on expected return and risk: the most promising projects compete for a chunk of half the funds, the "second tier" projects for a quarter, and so on. Projects with lower impact are lower tier, projects with lower probability of success are lower tier. We stop once we've rounded-up an allocation to the smallest unit of funding, hence avoiding Pascal's Wager.

Also note that there are only a finite number of options to choose from, because there are only a finite number of submitted research proposals.

Reply on news.ycombinator.com