4 ms·

It seems like a pretty strange null hypothesis that people are entirely unaffected by the social media they are exposed to. The fact that using Twitter bots has

by baddox 7y ago

It seems like a pretty strange null hypothesis that people are entirely unaffected by the social media they are exposed to. The fact that using Twitter bots hasn't caused one candidate to become the most popular doesn't seem like sufficient evidence. Perhaps he would be even less popular without the Twitter bots. And perhaps many of the other candidates are doing the same thing, but better or at a larger scale.

retsibsi 7y ago
It's also possible (and plausible to me) that negative messages tend to be more effective than positive ones, and that sowing doubt is easier than positively convincing people. In which case it would be very hard to elevate a random candidate into a winning position, but significantly easier to harm the reputation of a rival, or to muddy the waters on issues unfavourable to your candidate.
ipsum2 7y ago
> It seems like a pretty strange null hypothesis that people are entirely unaffected by the social media they are exposed to. Why? When people measure how ads perform, the null hypothesis is that the ad has no effect on the behavior of the viewer. Bots are sort of like ads that look like native content. That being said, I don't think a single data point (Yang) is enough to be conclusive, but it does change my beliefs a little bit on the effectiveness of bots.
- scarmig 7y ago
  Individual ads, sure. But an equivalent null hypothesis would be "social media advertising has no effect on people," which would seem pretty strange.
  - retsibsi 7y ago
    Also, the null hypothesis is just a starting point from which you gather and analyse the data that may disprove it. There's no reason that 'X has no effect on Y' should have a privileged position in messy real-world epistemology; we don't assume that nothing causes anything in the absence of conclusive proof, we keep an open mind and make our best guess based on whatever information we have.

by baddox 7y ago

retsibsi 7y ago

It's also possible (and plausible to me) that negative messages tend to be more effective than positive ones, and that sowing doubt is easier than positively convincing people. In which case it would be very hard to elevate a random candidate into a winning position, but significantly easier to harm the reputation of a rival, or to muddy the waters on issues unfavourable to your candidate.

ipsum2 7y ago

> It seems like a pretty strange null hypothesis that people are entirely unaffected by the social media they are exposed to. Why? When people measure how ads perform, the null hypothesis is that the ad has no effect on the behavior of the viewer. Bots are sort of like ads that look like native content. That being said, I don't think a single data point (Yang) is enough to be conclusive, but it does change my beliefs a little bit on the effectiveness of bots.

scarmig 7y ago
Individual ads, sure. But an equivalent null hypothesis would be "social media advertising has no effect on people," which would seem pretty strange.
- retsibsi 7y ago
  Also, the null hypothesis is just a starting point from which you gather and analyse the data that may disprove it. There's no reason that 'X has no effect on Y' should have a privileged position in messy real-world epistemology; we don't assume that nothing causes anything in the absence of conclusive proof, we keep an open mind and make our best guess based on whatever information we have.