Ricochet is the best place on the internet to discuss the issues of the day, either through commenting on posts or writing your own for our active and dynamic community in a fully moderated environment. In addition, the Ricochet Audio Network offers over 50 original podcasts with new episodes released every day.
Introducing: The Tone Analyzer
IBM Watson has sent out a press release, herewith dutifully reproduced by Gizmodo:
Taking a break from treating cancer and making cocktails, IBM’s Watson is now turning its attention to how people write. The supercomputer has been trained to judge the tone in people’s written messages—and can even give feedback about how to change it.
The IBM Watson Tone Analyzer sifts through passages of text to identify the emotional and social tone contained within, as well as the writing style. It can work out if the passages is cheerful, negative or angry, and then classify it depending on the openness, agreeableness, and conscientious of the message. Finally, it’ll describe how analytical, confident or tentative your writing style is.
That all allows you to tweak the text, with Watson making suggestions about which words can be changed to tweak the tone. The idea is to help guide people to write content that better suits their needs. Think of it, I suppose, like a spell check for tone.
I’m of several minds. On the one hand, I’m the one who said, “I bet we can automate the ‘cordial’ part of ‘cordial conservative conversation.'” Remember that? If Americans can buzz Pluto, they can automate the tone. So I feel vindicated at the thought that someone, somewhere, agrees with me.
On the other hand, since the beneficiary won’t be me, but IBM Watson, I’m completely against it. This thing will put me out of a job if it works.
On the third hand (that’s just my rhetorical forelimb, don’t stare), I’m not sure they’ve quite achieved what they’re hoping for:
What’s your verdict: Successful tone analysis? You think this gizmo’s a threat to my job?
Published in Culture, Entertainment, General, Science & Technology
I’m of an age and, as you know, in the field, such that this has personal poignancy. It was bad enough when Mom and Dad—elementary school teachers, and Dad eventually a national award-winning principal—sadly informed 8th-grade me they could no longer help me with my homework (Algebra I). It was much worse when 9th-grade me sold his mostly-unused drum set and bought a Model I TRS-80. No matter how I tried to bring them along, show them, explain to them, they didn’t get it. Very sad. But they were proud and happy when I was first published in High School, etc. One evening our phone rang. Dad answered it. “Yes. OK. Uh-huh. Right. Please excuse me, but I think you want to talk to my son.” Someone had gotten hold of our number, and had called with some abstruse question of TRS-80 arcana.
Cont’d.
When the Macintosh was introduced, it was a revelation. A machine my parents could use! I wasn’t able to buy one until 1985. It precipitated a crisis, with its OS of ~600 functions vs. any TRS-80 OS’ of 50-60, the Motorola 68000 and its sprawling instruction set and addressing modes compared to the humble workhorse Zilog Z-80. And GUI programming! Mac OS Classic was not Windows or Mac OS X, i.e. a GUI paint job slathered onto a fundamentally command-line OS. There was no command line. There was no character-generator ROM like on my Model I TRS-80, so I could open it up, solder in a 1-bit RAM chip, and suddenly see the lowercase that was always in the ROM, but the bit in the video RAM was missing that bit to save manufacturing costs. Every character was from a bitmap font (and, later, vector fonts we invented at Apple, and I code-reviewed the first renderer for, called “TrueType”).
I thought I was too stupid to be a programmer anymore.
Fast forward a year and a half, and I’m explaining to Andy Hertzfeld why the de-facto standard Mac debugger I work on doesn’t work on Macs with his 68020 daughterboard. Another couple of years, and I’m the #1 or #2 most requested MacDTS engineer in Apple’s debugging labs at MacWorld and WWDC.
Oh yeah: and informal tech support for Mom and Dad to this day.
Great Ghost of Gödel
Did you ever get to meet Doug Goodall?
I hate to say it, but: sometimes, yes.
We really need to get past our insistence on flat, 2D interaction technology. Datasets are too vast, and more importantly, much data in these vast datasets is represented in what are, mathematically, vector spaces of very high dimensionality (here’s where Midge’s pulse quickens…) You want rather advanced tools for manipulating such datasets mathematically, and at least as significantly, similarly advanced tools for presenting and interacting with the data. When I interviewed a few years back at one of LA’s ubiquitous Silicon Beach machine-learning/ad-targeting shops and we got talking about social media graphs and sentiment analysis and feature selection and so on, I tried to sell them on Dr. Munzner‘s work, for visualizing, analyzing, curating, training data, etc. They seemed intrigued. I never heard from them again. Recently I heard they got bought by AOL. Then AOL got bought by Verizon. I work for Verizon. So it all worked out in the end.
Darn it, G^3! Knock it off, willya?
I don’t.
You realizes this raises an obvious question.
No “the wine is agreeable, but the meat has spoiled” from you, eh?
I’ve never understood the “reach for the thesaurus” approach to editing, either. But that doesn’t affect the number of times I’ve seen it one iota.
I wonder if this is related to the work IBM is doing for the DoD on sentiment analysis and manipulation in social media?
Could be, but at this pace, I’d suggest they invest the money in more promising research.
It would have to be, in order to warrant using Watson for it at all. That is, they have a big data problem—likely in the petabyte range. “Tone analysis” of some reasonable degree on some reasonable corpus is well within the ambit of my 8-core 2.67 GHz Intel i7 processor 16G RAM MacBook Pro, especially if I put its 2G VRAM nVidia GeForce GT 750M GPU on the job (GPUs love linear algebra. You might even say they live for it).
As for manipulation, all sentiment analysis is for manipulation. 0% of it is for “well, that’s interesting. Thanks for the information!”
You can use shockingly naïve sentiment analysis on strictly public information sources to create market-beating securities returns today. The HFT systems actually used by the big traders, as well as very successful small ones, aren’t naïve. It’s important not to read too much into the dumbed-down public face of technology like this, where the operative phrase is “dumbed-down public.”
Claire,
Take chess as an example. A fixed little universe of 64 squares 2D. Five different major pieces and a pawn. Their movement clearly defined by strict rule. Winning the game simply and clearly defined. How difficult for an incredibly fast logic machine like a computer to solve this and defeat all feeble humans.
It took about 30 years of trying. The trying included custom hardware “chess position processors” and a long series of software improvements. Even then bringing down Grand Masters and finally the World Champion was touch and go.
Those darn Humans. Just a little more resourceful than you thought. Ain’t that a kick in the pants.
Regards,
Jim
Once again, I have to offer a caveat here: in contending with grand masters, for now, yes. But for the overwhelming majority of chess players, any of the COTS systems have provided more than enough challenge for those 30 years. And once there’s enough processing power, chess is over for humans, precisely because it’s a 2-person zero-sum game with perfect information.
The game that really resisted computerized attack and even required the invention of a new machine learning technique—temporal difference learning—was backgammon, with its roll of the dice and need to make the best of bad luck. However, once Gerry Tessuro invented temporal difference learning, it was over for backgammon, too: TD-Gammon learned to play by playing tens of thousands of games against itself, and its descendants are now trivially better than any human.
The real frontier remains Go, which makes chess look like checkers. We’ll see how it fares in another decade…
GGG,
Even though it would seem a lost cause (Grand Masters still hanging on against the Borg attack) and with a zero-sum game I don’t appear to have a leg to stand on (I must levitate), I still place my bet on the Humans. After all it’s been 45 years now surely the Humans deserve a little credit.
Humans can do the darndest things.
Regards,
Jim
The interesting thing is that with the advance of hardware and the discovery of various techniques it is now trivial for a human to write a chess program that is much stronger than the writer. (Obviously the degree of triviality varies with the programming and chess chops of the writer.) And yet deploying a complex system of programs that won’t leak personal data is beyond humanity.
Humans can, indeed, do the darndest things. It’s what they choose to do that confounds…
Writing a composing program that outputs better music than the programmer is maybe not so trivial, but also has been done – and not at the level of “even a tone-deaf programmer can program a computer to write a chorale”, either.
Indeed, I came back primarily to add “the significant thing about human grand masters isn’t that they’re human—it’s that they’re grand masters.” Chess is one of those domains where the gap between the top 1% and the merely brilliant seems immeasurable.
I suppose I would add: look at the gap between the current generation of ChessMaster and Deep Fritz. Now observe that Deep Fritz makes a big deal out of multicore support. Guess what? My chess program will suffer on my laptop, too, if it has to play with 7 of its 8 brains tied behind its back.
Gen,
Now we are getting somewhere. Humans have free choice. Humans have will. What they fix their will to and what choices they make are very revealing about their character. The corrupt and corrupting nature of the present Administration seems to permeate all levels. That sensitive data must be kept secure wasn’t a priority.
First maxim of this Administration, the government exists to provide jobs for government employees. That’s why they don’t want anything as shoddy as Obamacare for their own Health Care plan. That’s why if veterans are dying from inattention you don’t need to do anything about it. Second maxim of this Administration, anything that gets elected those who believe in the first maxim is justified. Thus an insane border policy that no other country in the World follows is absolutely necessary.
Now we must convince enough Humans to vote to throw the scoundrel out of office. Humans can be idiots but they can learn from their mistakes.
Regards,
Jim
GGG,
First, you tech tease you! Of course, I’m dying for a copy of Deep Fritz damn you. Even on a laptop its optimized for 64 bit Windows 8.1 and all that Web Site stuff is great too. (Like I actually have the spare time to do this you rascal)
Now as to what you were saying about Grand Masters it brings to mind Fischer. When I was on my high school chess team (no age jokes please) Bobby was at his peak ready to challenge the Russians. At one point he was ranked 3200 and the nearest Grand Master was ranked 2900. We made the joke that there would soon be a new classification system. Beginner, C-B-A Amateur, Expert, Master, Grand Master, and Fischer. He’d have his own category. What was most strange about Bobby was his motivation. He hated the Russians and wanted to crush them. The reason was obvious if you knew his background. He and his elder sister were abandoned by his communist parents who went off to fight the revolution or to just avoid responsibility. His elder sister got a job and supported them both. She bought him a $1.98 chess set as a gift when he was seven years old. He hated the Russians.
Got to watch out for those motivated Humans.
Regards,
Jim
Can it teach me how to tell morons where they can stuff their idiotic computer “problems” in a tone that makes them think I feel genuine sympathy and concern?
The demo does provide lots of opportunities for hilarity and wonder. I’ll give it that.