If polls are always wrong, who gains?

"One possible reason opinion and exit polls failed to anticipate the Bharatiya Janata Party (BJP)'s spectacular performance in the assembly election in Uttar Pradesh was that they couldn't quite figure out how to treat the 2014 general election," writes Karthik Shashidhar. In 2014 the BJP got 40% of the popular vote in UP which gave it 90% of seats in the Lok Sabha, whereas in the 2012 assembly elections it got just 15% of votes, which gave it 50 seats in the assembly. Except News 24 Chanakya, which gave the BJP 285 seats, all the other exit polls predicted a hung assembly. In fact, exit polls got all the 5 states wrong. In Punjab a poll of exit polls, which takes an average of all the polls, gave the Aam Aadmi Party 54 seats, it got just 20. In Goa it made BJP a winner with 18 seats, it came second to Congress with 13, while in Manipur it gave victory to the BJP with 25-31 seats, but it came second to the Congress with 21. So opinion and exit polls got almost all the results wrong. But this is not limited to India. In the presidential election in the US last year every opinion poll gave the election to Hillary Clinton, except USC/LA Times which consistently predicted a victory for Donald Trump. Exit polls predicted Clinton wins in all the vital battleground states of Wisconsin, North Carolina, Florida and Pennsylvania, but Trump won all of them. With just 50 out of a total of 403 seats in the old assembly the BJP should have concentrated on its strongholds to maximize its chances, writes Shashidhar. "Based on the data, however, it appears that the BJP has not followed this strategy. Instead, the party went for a broadbased strengthening across the state, and scored victories even in constituencies where its vote share was in the low single digits in 2012." Trump campaigned in and won counties won by Obama, majority of Clinton's votes came from the heavily populated Democrat states of California, Hawaii and New York.  Clinton got 2.9 million more popular votes than Trump, but could only manage to win 487 counties, less than Obama's 689, to Trump's 2626 counties. Instead of the 2.9 million if Clinton had just 100,000 more votes in key states she would have been president. Clinton's campaign team was so sure of winning Michigan that they did not pay attention to the state, Trump won Michigan by a mere 10,000 votes. Trump not only won the presidency but dragged down ballot Republicans to victory as well. Trump claimed that he campaigned for electoral college votes and would have campaigned differently for popular votes. The pollsters got it wrong in the US in 2016 and in India in 2017. Why? Perhaps, they are not polling a large enough sample across a majority of counties, because of the cost and difficulties involved. Still polls give television channels a chance to increase eyeballs and pundits a chance to show how clever they are. Even if they are always wrong.