I'm going to compare TA to weather forecasting. May not be the perfect example, but I'm not going to spend too much time trying to think of a better one at this time. Weather forecasting is not 100% accurate, and the longer the timeframe the less accurate it is. This is especially true when we consider winter snowstorms for example (for those of us who live in areas where it can snow). I can't count how many times snow was forecasted but when I woke up the next morning, the only snow I found was the dandruff on my pillow.
Thing is, every one still watches the weather forecasts because it is pretty accurate. But forecasts can change. So too with TA. It needs to be constantly monitored. I'm not going to speculate on the accuracy percentage of TA at this time. And, as others have said, I would say it is not as effective when applied to BTC as when applied to APPL for example. But yes, it is still effective when used on BTC. How do I know? Because I made a couple of trades that have made me money and have kept me from being in a losing position. That is why TAs use it, because it provides some kind of basis to forecast prices. I definitely would be clueless regarding BTC potential movements if it weren't for TA. Whenever I see large dumps that cause price to plummet, I'm not confused or surprised. In fact, I'm like "About time." I'm confident that the people who bash TA do so because they don't understand how to use it. It's not all about just lines on a chart either.
It's not the same at all. Not even remotely comparable.
Weather forecasting is running mathematical models through a simulator, looking for convergent outcomes, and assigning probabilities to them. It is mathematics based on science. It is not always accurate -- because that is the nature of probability; and the models can always be improved. The beef you have with weather forecasts may also be more to do with how they are presented rather than the forecast anyway. "48% chance of rain" is much more useful than a picture of rainclouds over your city.
On the other hand, technical analysis is based on the assumption that markets are predictable in behaviour due to regular patterns in aggregate human emotion. TAs will tell you that they are modelling market emotion. However they are not -- you will never see any such model presented. Rather they are using patterns in historical price and volume to predict future price and volume. Usually based on premises (such as Eliott Wave) that are 100 years old.
You will see lots of fancy indicators -- however they are all based on the same raw data: historical trades (There is no other data available). They are not even remotely modelling the underlying users and markets at play. You almost never see any TA that attempts to model the reasons for rises and falls. They treat price falls due to bad news, technical glitches, and manipulation equally, assuming that the average market participant knows nothing of the underlying causes too. This may have been (somewhat) useful for Wall St. a century ago, but to an Internet currency in 2013? Really?
Finally, another way that TA and weather forecasting is different... consider this. If TA could even remotely accurately model behaviour, the behaviour itself wouldn't happen and the price wouldn't move. This is the key as to why TA is effectively worthless. Is the TA influencing the market, or the other way around? Read around and you will soon realise it is the former. TAs are only market oracles to their herd of market sheep. Can't you see yourself how your last paragraph is testament to this fact?
Sorry to dig up an old post (just 1 day old actually. hah. this thread moves fast), but I really want to respond to this...
Your statement is full of sweeping generalizations
and some arguably true descriptions of how the math behind weather forecasts work and how the math behind TA, well, doesn't work. Which makes the argument sound more authoritative than it really is. Here's why:
You're picking the wrong subject to compare TA against (I know, you didn't really pick it, you responded). The more appropriate comparison would be
climate science.
Feedback loop: predictions influence behavior influence observations influence predictions? Check. Are the models still considered to give (within the set confidence intervals) accurate predictions? Sure.
Now, I'm not arguing that TA is on same the level of mathematical sophistication as climate models, but I wanted to point out that you can't really reject TA in its entirety because the possibility of feedback loops exist.
And another point: I personally find it much more accurate to think of "TA" as a subset of "algorithmic trading" (not in the sense that the word is usually used, but based on the pure meaning of a formal algorithm... doesn't matter if it runs on a computer or in your brain). There sure are a lot of terrible algorithms out there (I'm looking at you, Elliot waves and Fibonacci retracements), but you can't dismiss
all algorithms based on the existence of a few? many? maybe even
most algorithms being not useful.
But what you probably actually have in mind, and what I would agree with, is that where TA means "drawing pretty trend lines", it is almost completely useless.
Make note of my use of the words "maybe" and "potentially" and then, well, do whatever you like 
