561 karma · joined February 20, 2007
education is a good with almost-inelastic demand, hence
we can raise prices arbitrarily and in addition
eliminate government support for it :)
then they would not have been taken seriously.Wrt "rich", other posts indicate that $70K/yr household income as the upper limit for financial aid in CA. That's hardly "rich".
2. Define "able". Are you really advocating for a situation in which middle-class family routinely scrounge to save for college, or in which parents sell their houses to pay for college?(Btw, some countries (e.g., India) have a dowry system which requires parents of girls to save from the day the child is born and obviously it sucks to be those parents).
3. The original article seems to using the poor as a "human shield" to effect an arbitrary hike in fees. The reason the universities are not supported well is that govts are wasting money. I hope that resisting tuition fee increases will also pressure govt to reduce waste.
Assuming each listener played the song 100 times in 5 months (slightly less than once per day), we have 10,000 listeners. $167 for 10K listeners is 1.6 cents per listener (quite low compared to the 10euro/mo Spotify charges). But the number of listeners could easily be lower, if people listen many times to their favourite songs (like me).
I thought their goal was to eventually insert video ads into videos; that goal will not be impacted by people watching videos on TV.
And it can't be copyright concerns either.
I like the lala.com model the best. 10 cents to stream a song an unlimited number of times. No fixed monthly fees. Joined a couple of weeks ago; so far, I'm loving it. (If you want to "own your bits", you can get a DRM-free MP3 for an additional 79c).
Just yesterday I kicked off an MP3 playback using Rhythmbox on Ubuntu 9.04 and forgot to pause. When I came back a couple of hours later, two cores were pegged at 100% and the song was fast forwarding like crazy with no obvious UI control to slow it down.
And this is after wrestling with Adobe's 64-bit Flash plugin that doesn't play nice with USB audio (not sure if the 32-b version does; I'm not going to reinstall Ubuntu just for that).
This illustrates one risk for companies built on the idea of outsourcing infrastructure to Amazon AWS (Smugmug, Dropbox etc). You better add much more value on top of AWS, otherwise a Google/Microsoft can undercut you big-time on price.
My perception is that they haven't dropped prices enough although hardware has become much better (1 GHz equivalent machine as baseline?, I built a quad-core machine for $600 and got a 1TB drive for less than $100).
Google's move definitely validates my hunch.
I was thinking about getting a Smugmug account, but now I will probably use a Google account.
Building fault-tolerance (like S3/EBS do) is a HARD problem. That's worth paying for.
Once you move your data to the cloud for reliability, you almost have to move the processing close to the data to get acceptable performance.
That's what interests me the most.
The stuff they have here for annotating types (love the type expressions), equivalent of #define are major conveniences to ease development. And long overdue. So thanks, Google.
I found out about SIN/COS a while before I really understood what they were. :) I treated them as magic keywords and used to just plug in weird functions to see what turned out (sin(x)+cos(x), sin(cos(x)) etc).
Sometimes we take things like portability and platform-independence so far that we end up complicating things. As an example, with Turbo Pascal and DOS, I could just treat the video buffer as an array and write to it. Under Linux (even in the old days), I wasn't able to figure out how to access the console display as a array. The "right way" was to wrestle with Curses. I declined the invitation and stuck to network/command line programs. :)
Seems like they're stretching the definition of bug here. Maybe a bug in the specification, almost certainly not in the code.
But tmux has another problem: very rapid scrolling in one window will make the program completely unresponsive. In the same situation, screen doesn't have this problem. I posted to the user list and the developer very quickly sent me a patch, but it didn't fix the problem for me, although it worked for the developer. Still working with the developer on this.
What appears to be a long jump to a non-expert on first glance, on deeper inspection will usually turn out to have involved multiple small incremental steps culminating a final discovery. Since we have to acknowledge somebody we usually pick the guy/gal who finished the job. I'm not saying that all work is equal, but imho planning for long jumps is impossible (it's not like researchers don't want to make big leaps in understanding).
Examples which have changed my mind about this: Andrew Wiles' Fermat proof, evolution of RISC, quantum theories. I am pretty sure that if I dig into it, I'll find that relativity and evolution also built upon previous work.
Many of the programmers repeatedly emphasized the importance of data structures to both understanding programs as well to solving the problem. And after noticing how I work, I find that to be true.
English documentation of the core data structures is more important than verbose description of the steps in the algorithm. This is the key thing that is different among progarms. By and large, we use the same algorithms again and again. And I can trivially find English documentation of those to understand rapidly. I don't need it in the code also.
And by documentation I mean something that explains at a programming language level with the identifiers etc. For git, it would be insufficient to say "trees are collections of blobs". I would need to know the struct names and an explanation of each of the members. And whether the "collection" is a array/list/hash etc.
Having looked into literate programming, it seems to emphasize documenting the algorithms more than it does the data structures. Hence it misses the "big picture" which is what is really needed.
I'm not convinced that LP is useful.
How about ulimit? Could that be useful?
Maybe they propose making the instruction count part of a task's "context". My understanding is that today's chip instruction counters are per-chip and not per-task.