Edward Chancellor's Devil Take the Hindmost: A History of Financial Speculation is a history of several market bubbles and crashes. It covers bubbles such as the South Sea Bubble, the 1920s bubble in the US stock market preceeding the great depression, the dotcom bubble of the 1990's and Japan in the 1980's. The main lessons I took was if a market sounds too good to be true it probably is, that highly leveraged financial instruments tend to prolong and worsen bubbles and often the people who bear the cost of reckless speculation are different to the people who take and profit from it.
The flow of the different chapters was haphazard; the earlier bubble seem to be only included to illustrate recurring themes in the later bubbles which are discussed in too much detail with too many footnotes. The book is really informative and worth wading through, although you may get many of the same themes more enjoyably from reading Michael Lewis' books on the financial industry.
A recurring theme was that markets that seem guaranteed to rise are almost certainly a bubble. Many times contemporaries said the traditional rules of valuation didn't apply to them because of technological or financial progress, or in the case of Japan because of cultural differences. Similar cries are heard today of how automation will replace all workers, how people will soon live forever, or that the artificial intelligence will lead to a technological singularity. Because of these, or more pedestrian reasons, it's claimed we can't apply historical trends into the future. However it's much more likely that present technological advances are similar to advances in the past, than that we've reached some new revolutionary point where previous rules don't apply.
All the bubbles in the book are attributed to speculation; people buying things because they think the value will increase because other people will buy them. When people from all backgrounds start getting interested in something and think that they can't lose it's likely a bubble. The recent cryptocurrency boom is an example of this; growth was based on people speculating that it would grow more. However clearly there's a ceiling to the value of things; if the price is much higher than the value of the underlying asset (judged for example by earnings or assets or dividends) then it's likely something is wrong and won't be sustained. It's really hard to know if you're in a bubble; I strongly suspect the property markets of Melbourne and Sydney are overvalued (based on price relative to rental yields and income) but regulation and market conditions means this has been sustained for a very long time.
A hard fact was often the speculators can profit greatly, but cause great harm. Sometimes they are personally bankrupted, but especially recently they are often protected by incorportation and the government. An example was the Savings and Loans scandals, where they speculated with customer market covered by a US government guarantee, so the taxpayer ended up footing the bill. If a company would be bailed out they should be heavily regulated to ensure they're not exposing the government to unreasonable risk.
After reading this I would stay away from new and complex financial products. The longer the history of a product and simpler to understand the less likely it will collapse unexpectedly. Products like junk bonds and derivatives have shown a lot of great harm; while you want to invest in something to outrun inflation, using an old investment vehicle makes it more likely you know what you're getting.