Tesla So I don't have a car and don't plan to buy a car any time soon (I live a mile from the office and sometimes walk to work) but I find electric cars to be really really cool.
If I did buy a car, I would give serious consideration to getting an electric car as my sole vehicle if that were at all practical (if for some reason I needed to buy two cars the second car would almost definitely be electric if the first one was gasoline powered)
So recently I stopped by the Tesla Store in DC just to check it out and had a pretty long chat with a salesperson there named Shaun. It almost left me wishing that I lived in Woodbridge, Virginia (i.e. somewhere really, really far away) so I could justify thinking about reserving a Tesla Model S (this is a $50,000 Luxury Sedan that will ship in June 2012). OK, I probably still wouldn't buy a $50,000 car at this point in my life, but I would think about it for a while ;-)
Here are a few things I learned:
(1) Tesla cars have ANOTHER trunk beneath the front bonnet since the electric motor is much less bulky than an internal combustion(IC) engine. This essentially translates into TWICE the trunk space.
(2) The lack of a drive shaft makes a noticeable difference as it gives you a nice flat chassis which in turn leads to a roomier interior.
(3) Since electric motors don't have the constant grinding friction of an IC engine so they last a lot longer with little service. The batteries also have longevity but they do lose some of their capacity over time. But Shaun claimed the batteries still have 70% resale value after 10 years so a 20,000$ battery could be traded in for a new one with the addition of $6,000$
(4) There is some very interesting research to be done with battery technology. It seems that the main bottleneck in getting faster recharges is that Lithium Ion batteries will quickly lose their charging capacity if they are recharged too fast. So even though high capacity charging stations exist they recommend you shouldn't use them too much or you battery will deteriorate. Tesla considers reducing recharge time their most important research challenge (i.e. even more so than increasing battery capacity).
(5) There is a Moore's Law of battery technology where the battery capacity seems to double every 7 years or so. So in 2019 perhaps every electric car will have twice the driving range it has now. I personally think a range of 600 miles on a single charge will be a sweet spot for me because you could drive from DC to Pittsburgh or DC to NY without any sort of range anxiety or being forced to stop. Actually 300 miles is about how far you can go on a full tank for many gasoline cars now that I think about it.
So I hope either Zipcar or Enterprise Rent-A-Car (my two main sources of cars) widely offer electric cars as options in the next couple of years as I would definitely go for that (even though they would probably be more expensive than what I would typically get).
A Saturday Run I JUST finished reading Haruki Murakami's "What I Talk About When I Talk About Running" which I liked a lot, but I am not sure non-runners would be able to appreciate it. He writes about running, growing older, what running taught him about writing and about himself. Immediately after reading it I felt inspired to chronicle a run I did yesterday. It seems these kind of experiences are fairly typical when running.
My typical pace is 12:00 but for some reason I decided to go with the 10:30 pace group. There probably was some self-delusion involved - just the other day I had done 5 miles at an 11:00 pace, surely a 10:30 pace was not out of reach? I started off quite well, in fact it felt too easy. We passed the Georgetown waterfront, briefly went on the Rock Creek Park trail and swept by the Kennedy Center and Watergate complex. Before I started biking in the area in summer 2010 I always felt rather embarassed by my lack of local knowledge. I am basically a homebody and rarely have personal or professional reasons to visit most parts of DC. But after running and biking for almost two years the area around Georgetown and The Mall felt very comfortingly familiar.
Can I say something about people who talk while running? For me this has never felt comfortable - perhaps I don't breath correctly. I guess my philosophy is that if you're running you should focus on running - this is not the time for chit chat! But obviously many people feel more comfortable chattering away - they probably find it to be a good distraction. I can only manage it if I greatly reduce my pace, if I am trying to actually run properly maintaining a conversation is out of the question.
Around mile 5 is when things broke down. I had started falling slightly behind my group around mile 4 but I was determined to keep going until the half-way mark of the 10 mile run. Finally I stopped for a walk near the Library of Congress. BIG MISTAKE. If there is one piece of advice I can give it's this: DON'T. STOP. EVER. I know some people who delibarately plan to periodically stop for breaks and it work's for them but I don't think I could ever do that. I have the kind of personality where once I lose momentum I am completely finished. And it's the same way in my non-running life - which isn't necessarily a good thing.
The next hour was pure hell. EVERYONE passed me. Even the girls who were running for the first time. (Yes, it's true some WOMEN are incredibly fast but that doesn't mean one enjoys being passed by them!) I tried to get into a system of walking for 2 minutes and running for 3 minutes but even that proved too strenous. I wasn't physically hurt or cramped - just fundamentally unsound. Eventually even the second group of runners - who started 30 minutes after we did - all passed me. The humiliation was complete.
Finally we reached the Key Bridge - the finish line was at the end. I am one of those runners who is good at giving a "kick" at the end - I try to hold something in reserve. So I thought I would try to sprint across this bridge(around 500-1000 feet). The attempt lasted 15 seconds.
My eventual time was 126 minutes for 10 miles which gives a dissappointing 12:36 minutes per mile pace. But considering I had ran the first 5 miles in 50 minutes, this means the second half was at a truly miserable 15:12 pace. If I ran an official race at that pace I would be disqualified.
When I got back to the Georgetown Running Company and was sipping on the Gatorade a girl who had passed me straggled in. I asked her where she was and she claimed that according to her Fancy GPS Watch I had actually only ran 9.5 miles and we were supposed to go a several more blocks to make up the difference. Damn!
When I recorded the run in my training log I counted it as 10 miles. Maybe her Fancy GPS Watch is broken.
"Steve Jobs" - Walter Isaacson I was worried whether there would be anything really new in this biography, since I thought I already knew quite a lot about Steve Jobs. Thankfully this was not a problem as there are lots of details that I hadn't seen in the numerous articles that have appeared recently.
The part that fascinated me most was the description of what it was like growing up in Silicon Valley in the late 60s to early 70s. There was a sense of great possibility and an inkling that the emerging computer industry would change everything. It's clear that if Jobs had grown up anywhere else his life might have turned out very different.
Like most people reading this book what I really wanted to learn about was what Jobs was like as a person. And I think I got a decent answer. The phrase that popped up that seemed most satisfactory was "Narcissistic personality disorder." He could be incredibly rude - even to complete strangers. He believed the normal rules shouldn't apply to him - which is why he always parked in the handicapped spot. Nobody who knew him would describe him as "Steve Jobs - a really nice guy."
But it is only fair to note that most people who worked for him seemed to appreciated him, that his demanding nature made them do their best work. And he had a lot of friends, so he wasn't just a random jerk all the time. He just wasn't the warm and cuddly type. He seems like one of those friends you have who is fun to hand around with but isn't a very good listener and might occasionally say something hurtful (and then apologize the next day).
The other thread that struck me was the strong thread of mysticism in Jobs' thinking. In simple terms he believed rationality and logical thinking were overrated. Hence his lifelong fascination with East Asian religions, his puritanical diets (in college he would go for weeks at a time eating only carrots) and his initial refusal to get surgery for his pancreatic cancer. This minimalistic, ascetic, pursuit of purity obviously deeply informed Apple products and designs.
So what could an average person learn from all this? Well in one sense - nothing. Steve Jobs was a unique individual. You couldn't act like him and expect to get the same results - unless you had the same personal make up.
But in another sense - Jobs was clearly very passionate about what ever he did. He only did exactly what he felt passionate about and didn't do anything he didn't feel like doing. This seems vaguely impossible but apparently he was very successful while doing so.
Not surprised that students cheat, as a rule if there is a way for students to get a good grade without doing too much hard work - a lot of them will take it. In most cases this involves copying the solution(s) from a classmate or from the web or other resource when it is explicitly forbidden.
The main point is that I agree with Panos that the best way to deal with this is not to spend a lot of time scrutinizing students for cheating but to to try to make "cheat proof assessment." i.e to make assignments where cheating is hard or where it really doesn't matter if the students "cheat."
This can be done in a number of ways and Panos outlines a few: (1) Make the assignments public, (2) have the students grade each other
and so on.
Here is what I ended up doing in my classes (which are mainly math and programming courses like Analysis of Algorithms and Machine Learning): (1) Freely allow collaboration and Googling on homework assignments. That way they CAN'T cheat even if they wanted to :-) (2) But make sure homeworks don't count for very much of the overall grade.(most of the grades comes from in-class quizzes and tests).
This way I am free to make the homeworks fairly challenging and even if a lot of the students copy, there is a good chance that at least SOME students will actually work through the problem as intended and really learn something.
This wouldn't work for every class (esp. a class which requires essays or projects) but it seems to more or less work now.
But the thing which really struck me was the strong reaction to Panos' post. Clearly people feel strongly about the issue of cheating. Interestingly enough ALL the commenters claimed to be people who never cheated who were disgusted by the prevalence of cheating. Some attacked Panos for saying that he was giving up on trying to catch cheaters (ignoring his stated plan to instead deal with the issue by making his assignments "cheat proof").
His story was discussed on several prominent sites (some of them non-academic):
I guess its good that people are so passionate about this. But from a teaching perspective the more important issue is: How can you assess that students are successfully acquiring the knowledge and insights that you want them to acquire? I think most teachers would be upset if they learned that someone could "pass" their class without acquiring any of the skills or knowledge they were supposed to be getting.
And I think this is only going to get more challenging as technology makes "cheating" easier and easier. When I ask a question in class a lot of times they will just look it up on their iPhones rather than ponder how to work out the answer. This is not cheating off course but is illustrative of the broader issue.
But of course I think the technology which has put much of the world's knowledge at student's fingertips is a great boon for those students who really want to learn something. But it has also certainly made "cheating" - however you define it - a lot easier than ever before.
Black History #Math Suppose you are working at the check out counter at your local supermarket.
You know that roughly the SAME number of customers arrive over any interval of time (i.e maybe it's 20 customers per hour or 60 customers per hour, you're not sure).
You notice the following number of customers arriving each minute over a particular 10 minute period: 1,4,2,3,2,4,1,0,2,1
You're interested in the following: What is the PROBABILITY of ZERO customers arriving in a 1 minute interval? (Perhaps you want to take a VERY short break!)
HOW TO CALCULATE!??? . . . After thinking about it you come up with the following EXTREMELY CRUDE way of estimating this unknown probability:
THE CRUDE ESTIMATOR:(i) if the number of customers in the FIRST minute is 0 then we estimate the probability of ZERO customers as 1 (ii) if the number of customers in the FIRST minute is NOT ZERO then we estimate that the probability of ZERO customers is 0.
For the numbers cited above we have 1 customer in the first minute so we would estimate the probability of ZERO customers as ZERO. . . . Obviously a HORRIBLE estimator right?
THE PUNCHLINE =>There is an easy "automatic" way to convert this horrible/crude no good estimator into a great/"perfect" estimator!
It turns out that after doing this we end up with (1 - 1/n)S as our estimator where n is the number of 1 minute intervals(10 in our example) and S is the total number of customers over the entire period (20 in this case).
So the probability of zero customers in a 1 minute interval is (1-(1/10))^20 = 0.12
He completed his PhD at University of Illinois at Urbana Champaign in 1941 at age 22
In 1941-42 he was denied use of the facilities at Princeton University while a post-doctoral scholar at IAS because of his race.
Around 1942 the great statistician Jerzy Neymann wanted to offer him a position at Berkeley but "race-based objections prevented his appointment at the time."
When seeking a permanent position he applied to all 105 Historically Black Colleges as "he felt at the time that a black teacher would be limited to teaching only at black colleges"
Fortunately Prof Blackwell's tale has a happy ending
In 1944 he obtained a position at Howard University. Within THREE years he was appointed FULL professor and Chair of the Department
Despite a heavy teaching load, his time at Howard was extraordinarily productive resulting in many ground breaking publications in statistics and game theory (including the one on the Rao-Blackwell theorem)
In 1954 he took a job at Berkeley and soon became Chair of the Statistics Department. He stayed there until his retirement in 1988 after a long and fruitful career filled with many honors and awards. He also married and raised 8 kids ... and mentored more than 60 PhD students.
Here is something a lot of reviews didn't comment on: The epilogue of "The Last Ring Bearer" explicitly casts doubts on the entire preceding narrative, suggesting "The Last Ring Bearer" may be just as biased towards Sauron as "The Lord of the Rings" is biased against him. So in this sense it has a thoroughly post modern sensibility in contrast to the more romantic outlook of "The Lord of the Rings."
At any rate, I thoroughly enjoyed it and the motivations of the characters seemed to make more sense in a lot of places than in "The Lord of the Rings."
Thoughts on Watson (1) AMAZING: I think it is absolutely an AMAZING achievement. We have now built a computer that can answer factual questions stated in normal human language just about as quickly and accurately as the best humans. That is pretty great.
(2) THE ENSEMBLE: When Watson receives a question it fires up about a 100 different algorithms which each tackle the question and come up with their own answer(s). Some of the algorithms specialize in geography, others in literature, others in puns, others in history and so on. These answers are then evaluated and the best ones selected. This kind of technique of combining several different algorithms is sometimes called "Ensemble Learning" in the machine learning literature. It is interesting that the algorithm which eventually won the netflix prize also used an ensemble approach. Actually ensemble learning is nothing new, but it is interesting how effective it can be in practise.
And what is really intriguing is the possible connection with how the human brain works. We know that our brains are surprisingly specialized. There are regions of the brain which deal only with irregular verbs(!) for example. And of course we have specialized regions for vision, hearing, motor skills and so forth. So why couldn't there be a "talking to my mother in law" region or a "giving machine learning lectures" region? It seems pretty plausible that when we are confronted with a situation different regions of the brain chime in and some kind of "vote" takes place to decide what to do. This appears to be exactly the sort of model Marvin Minsky was arguing for in "Society of Mind"
(3) "More Data Usually Beats Better Algorithms?": Clearly there were advances in algorithms and hardware in Watson. But it seems a big part of Watson's success is simply being able to access sufficiently large amounts of data (15 Trillion bytes supposedly). I also think of how Google Translate became so good after scarfing down the 200 billion words of the UN parallel corpus.
(4) Some Random Applications
(i) Periodically search the web and alert me to any papers I should read or proposals I should submit. Something like that already exists, but you have to imagine Watson would be more accurate.
(ii) Combine with voice recognition for natural language queries - e.g. [spoken aloud] "Computer: How much is a DC to Jamaica round trip this weekend" (yes you could find the answer with 5 minutes of googling, but why waste 5 minutes?)
(ii) Watson, the scholar?Richard Lipton has some interesting ideas on how something like this could be useful in research. If nothing else the natural language processing should make it easier to find all the relevant work.
One phrase that jumped out at me was "Can handle high degrees of ambiguity or uncertainty"
That seems like something that is applicable to academia, since you never know which grant is going to get funded or which long-shot idea is actually going to pan out, or what you will be working on next year, etc.
OK, so maybe academia is not very unique in this respect, but I thought the article was a good read in any case.
This Weeks Links #1 (1/24 - 1/30) Twitter is great in a lot of ways, but it is not really that good if you want to go back and read a link that was interesting from several months ago. Blogs are at least slightly better in this respect, so I decided it might be a good idea to collect the best links from the past week into This Weeks Links.(Yes, the title is a deliberate homage to the great John Baez's This Weeks Find)
I actually didn't realize what was going on until the end, but I think it speaks for itself.
Very accessible talk by Prof. Ken Ono of Emory on the recent breakthroughs in understanding partition numbers. I literally didn't know what a partition number was before I watched this video, but still followed almost everything (at least in the first easy half before he starts banging on about fractals).
Last Sunday I visited the Computer History Museum in Mountain View. I really had a good time - the exhibits are vast and comprehensive. I was particularly fascinated by the operation of the Difference Engine - the vast swaying spirals of its shafts as it calculates are nothing short of hypnotic. Seriously!
Pretty simple concept, but this article really hit me as I also have this problem of "waiting to be struck by inspiration." But often when I just start doing something sans inspiration, something else comes up and then something else - and suddenly things are getting interesting! But just because I know that intellectually doesn't mean I always put it into practice...