Brad Templeton is an EFF
director, Singularity U
faculty, software architect and internet entrepreneur, robotic car strategist, futurist lecturer, hobby photographer and Burning Man artist.
This is an "ideas" blog rather than a "cool thing I saw today" blog. Many of the items are not topical. If you like what you read, I recommend you also browse back in the archives, starting with the best of blog section. It also has various "topic" and "tag" sections (see menu on right) and some are sub blogs like Robocars, photography and Going Green. Try my home page for more info and contact data.
Submitted by brad on Sun, 2005-01-02 12:08.
Just back from the nightmare of holiday travel, which started at 5:30am on Christmas morning and a security line snaking all the way to baggage claim. Coming back 6 days later, I braved the door to door shuttles from the airport.
I generally regret the decision to use these shuttles, which seem to average about 1 hour 30 minutes for the 35 minute drive to my home from SFO. This time, they had 10 people waiting for my town (which would normally be a dream as you would not spend all that time wanding around closer towns dropping off earlier folks) but in fact after we saw others had waited an hour for any shuttle to show up, we went to the caltrain, which takes an hour for the trip but is predictable.
The curse of these shuttles is how unpredictable they are. For some they are a quick trip but often they will drive you many times around the airport waiting for passengers, and then on an unpredictable drive. The public hates unpredictability even more than slowness, and would pay for predictability, I think.
So can computers, and some common sense, fix this? Surely you could make reservations which tie your flight number into the database so the shuttle company sees your plane arrive and knows pretty accurately when you will make the curb. (You can confirm that with a cell phone speed dial if your cell number is registered.) If lots of people did this, you could know how quickly a large enough group of people who live close together would be ready to leave. read more »
Submitted by brad on Wed, 2004-12-15 06:33.
As I continue to play with HDTV, I found I had a horrendous time getting good output from my computer running the MythTV open source PVR into my TV. DVI, the uncompressed digital standard, just wouldn't work from the video card I had to the TV. The TV has Firewire/1394, which would allow me to stream mpeg-2 to it, and that would be really great, but as yet no software supports it because few TVs have such inputs.
Here's another idea, one that reverses old thinking. The earliest VCRs did their outut on an RF modulator to channel 3, the only way to get into older TVs. Now we of course recommend non-RF methods, such as composite video, S-video or best of all component video/VGA, or in the digital realm, DVI and HDMI.
But in fact with TVs being mandated to have ATSC tuners, could it make sense to go back to RF? This gives digital decoding in the TV, in theory the highest quality if the TV has a good decoder. The cables are easy and super cheap, and carry digital video and 6 channel sound -- something even DVI doesn't do. You can run them as long as you want.
Plus, you can have tons of multiple inputs, just on different channels. Put your cable box on channel 3, your PVR on channel 4, DVD player on channel 5 and so on -- no need for the plethora of inputs and mass of cables.
(Don't get me wrong, I think a single ethernet jack would be better than any of these methods, but the TVs don't have them and they do have the tuners now.)
There is one big issue, however, which is on-screen display and comptuer generated menus. The RF sends a compressed mpeg stream. On the
surface, that's great because the boxes handling the video can be slow and
cheap -- they are just slinging bits they don't understand. But once you want
to overlay text on the video, you suddenly have to decode the stream (hard enough) and then re-encode it, which is close to impossible with today's hardware. On the other hand, it should be possible to do non-transparent overlays, where you take over a region of the screen (perhaps the bottom is easiest) and replace it with your generated text.
The ideal solution to this would be to modify the protocols to allow sending a second stream to be overlayed, with an alpha channel, on the main one. This is true no matter how you send compressed video -- RF, ethernet or firewire. However, we don't get to change the protocols, the idea here is to make use of something already out there.
Generating Mpeg from computer menu displays on the other hand is something that should be within the capabilities of today's CPUs. They can do it at smaller resolution (720x480 DVD res is fine) but more to the point unless there are fancy animations, it's static and machine generated and easy to set up for quick conversion.
There's no encryption, which might cause pressure to balk at this but it has a lot of advantages worth considering as kludges go.
Submitted by brad on Wed, 2004-12-15 05:18.
In my quest over the leak/sale of the entertainment.com mailing list, I have some amusing updates.
After telling them you don't respond to a "You sold my name" complaint with a request for all of the person's personal information, I got back yet another stock message, "Here's how you can get off our mailing list." I'm getting a lot of companies who use customer service reps for E-mail who clearly never read the E-mails. Yes, I also get software that auto-responds, but amazingly we also get humans who auto-respond.
Anyway, customer service clearly not working, I found their phone number and called their legal dept. where I spoke with a Jill Silverman. She expressed concern after she got clear on what had happened, and asked me to forward her the emails I had gotten to my special address created just for them. I immediately sent them off then heard nothing.
When I inquired again, she told me she never got the E-mails. I figured out why, eventually. The E-mails, which I had put in a text file attachment, were of course spams, and triggered her company's spam filter. Of course, my mail was dropped on the floor, no diagnostic for her or me.
So I put them in a web page and sent her the URL. That should make it through!
Submitted by brad on Mon, 2004-12-13 08:52.
Creationists regularly complain that schools teach evolution improperly and should also offer creation science as an alternative. They went so far as to push one school board to put stickers on biology textbooks remindng students that evolution is a theory and should be critically viewed.
Well, surprisingly, I have some agreement with them. Evolution, like Quantum Mechanics, gravity and others is indeed a theory. And in proper science all theories are subject to intense scrutiny and testing. They are required to make predictions which can turn out false, and those predictions are tested with repeatable experimentation and observation.
So now I wonder, why if we give them their way — sort of — and mandate the teaching of “creation science” in the shools. Except I mean a rigourous, scientific treatment, by non-religious teachers, where a lesson about science and bad science is taught. Other examples of bad science should also be covered.
Students should be challenged to consider the predictions, past and present, of the creation “scientists” and whether they have come true. They should learn what happens when people conclude the results in advance and try to bend the facts to fit them. It happens in all areas of science, and a good education trains you to identify when it is happening, and when you are doing it yourself. They should of course also learn the predictions of evolution and many other theories and how they have been tested and verified. They should learn about theories that had supporters but then failed their tests and thus fell from favour.
Why creation science and not every other bogus fake science? Well, studies show it is probably the one most widely believed by the public, though psychic powers, alien abductions and others also rank highly. So as the #1 it deserves a place in our curriculum, because the critical examination of bad science deserves a place.
Indeed, for a student not actually going into science, it could well be that learning to understand bad science would be the most important thing they take out of the program. They will almost assuredly never need to calculate the velocity of a spherical monkey hanging from a massless rope over a frictionless pulley. But they will encounter bad science and have to deal with it.
(I think the same is true in math for non-professionals. One of the most important things they should learn is how statistics are misused.)
So give them what they want, and then see them beg to take it back.
Submitted by brad on Wed, 2004-12-08 08:51.
When I give an E-mail address to a web site, I give a different one to each site. I have many domains, including one where all addresses are forwarded to me unless I turn them off.
Submitted by brad on Thu, 2004-12-02 07:25.
I have been building an HDTV PVR with MythTV and the pcHDTV tuner
card. It's been a major adventure, not yet ready for prime time, but
it's lead me to have some thoughts about things you want to think about
in a PVR that particularly relate to HDTV.
Suggesting new features is of course a somewhat futile activity. In open
source, the usually and appropriate answer is "why don't you go code up
this feature and add it?" In commercial products, most people feel
even the Tivo is too complex and they are correctly loathe to add
new features that complicate the user interface. So I make a priority
note on all of these.
If you are not familiar with certain linux video issues, some of this
will sound like gibberish. read more »
Submitted by brad on Wed, 2004-11-24 20:45.
For people of my generation, a great deal of history was seen on regular low-resolution TV. But a lot of it, up to the 70s or so (and often after) was shot on film, at higher resolution. Older generations saw some of this (once or twice) in newsreels at the movies.
So as HD sets become common, it would be great to see this old film footage of events like the wars, the Olympics, famous speeches, the moon landing and other space program material in high definition. I saw a DVD of the 1936 Olympics on a good screen and even that was surprising.
Some of this is already happening. HDNet is digitizing old sitcoms like Hogan’s Heroes, which was shot on film. Fun, but I think the life-changing events we’ve never truly seen (or rarely seen) at full resolution would be more important, and would also leave a legacy for education. And yes, that even includes that bad news, like the Challenger and the WTC.
Update: Discovery channel has an upcoming series of NASA footage coming up in HD. Now if only I could get discovery channel HD in a way that I can record without buying their proprietary box.
Submitted by brad on Wed, 2004-11-24 06:16.
A short note, as I've been busy with a number of things including trying out PC based HDTV recording and mythtv, which I will write about shortly.
In a VoIP pricing debate, I calculated an interesting observation. Today we have voice codecs that can do very fine voice quality in about 25 kilobits. FM radio quality. Cell phone quality can be done in 7 kilobits. Anyway, that means all 200 million adult Americans could be on the phone at once and it would use 5 terabits.
Using DWDM -- dense wave division multiplexing -- where many colours of photon co-exist in an optical fiber, 5 terabits is now well within the range of a single optical fiber.
So yes, everybody could be on the phone at once, in FM quality, with just one optical fiber. Of course one fiber doesn't pass by every town, and multiplexing it all together costs money too and has overhead. But it should be clear just how silly the idea of per minute charges for voice are in the face of this statistic. And indeed they are going away.
Of course we are finding uses for more bandwidth. HDTV file transfers for example (which use 16 megabits but can be credibly done in about 6 with better codecs.) But how long before we can all have an HDTV videophone call at once over a moderate number of fibers?
Submitted by brad on Tue, 2004-11-16 08:05.
This is a science-fictional idea, but strikingly probable. You are probably aware the brain is split into two halves, joined by a nerve bundle called the corpus callosum. People with severe epilepsy have had the callosum severed, and ended up having two brains in one body. A left brain controlling the right half, and a right brain controlling the left half. The left brain can speak and can lie, the right brain can write with the left hand to communicate. Until experimenters tried to communicate with each half, the people were unaware they were sharing their body with another half-brain. You can read more details if you were unaware.
Anyway, on to the idea. It seems plausible one could apply a temporary anesthetic to the corpus callosum, and temporarily split a person into two brains. Today that might require drastic steps like brain surgery. In the future it's not hard to imagine a specialized drug or highly targetted drug delivery or nanobots to temporarily numb and disable the zone, without too much shutdown of adjacent tissue.
One could even imagine recreational use if it were safe and simple. It would be quite something to learn just what happens when you are split into two and then re-integrated, and how the reformed whole you would remember the two split experiences. Would your left and right half have a conversation? What would they talk about? How long could you persist in this state before it became difficult to reintegrate? How long after the cutoff would you be cogent enough to do all this (one presumes the cutoff might be a bit of a shock.)
Imagine an SF story where one half of the protagonist's brain decides it doesn't want to re-integrate with the other half, and takes steps to assure this...
Submitted by brad on Sun, 2004-11-14 14:13.
There are a number of Linix "Live CD" distributions out there. These allow you to boot Linux from a CD and run it (somewhat slowly) without ever touching the hard disk in the machine. (They can access the disk however, which makes them good for system repair, both for Linux and Windows.) One popular one is Knoppix, and Mandrake makes one called MandrakeMove, which takes the important next step of letting you store your personal config choices on a USB thumbdrive or floppy. There are distributions that can fit on a thumbrive (after all, those drives are getting quite large for little money, but this is recent enough there hasn't been as much focus on this.)
Let me suggest where I would like this trend to continue. It's great to be able to take any machine and quickly convert it to your style and environment with a CD, or even better a business-card CD or thumbdrive. (Most systems can boot from a CD, fewer from a thumb-drive, most from a floppy leading to another device.) Storing some state on a floppy, thumbdrive, CD-R session -- preferences, home directory files and scripts, browswer config and bookmarks -- is a must. Indeed, if the tools let you build a custom CD just for you with your choice of packages you can bring in much of your whole working environment.
I haven't seen anybody provide automatic storage on the net, based on the assumption the machine you take over probably has an ethernet card. If it does, it would be great to go out and suck down your latest personal changes and files, starting with the most important to get you going, and bringing the rest in the background. This doesn't need a special server, though the group making this distribution might well offer to do so. You could keep and update much of this data in a special mailbox message or mailbox folder, especially with IMAP. Anybody can get access to that. (Or a web mail tool like GMAIL.) Of course if you have actual hosting this can also be used. The data would be encrypted, you would need a password -- not just your mail password -- to use it.
As you changed the data it would be updated to the net storage. Now you could go to any machine with a non-customized CD. Indeed, you could even, on a common fast machine, download a minimal environment (perhaps 60 megabytes which is just a few minutes on a fast broadband link) and after it boots, get the custom information including which other packages are important.
The key is to do things in the Windows filesystem, most likely to be what you find on the machine where you are the guest. read more »
Submitted by brad on Mon, 2004-11-08 12:56.
David Brin, whom I debated on the topic of Transparency yesterday, has been putting forward for some time the general idea of a henchman's amnesty law. Namely that, in the event of a criminal conspiracy, the first underling who whistle-blows can get some level of amnesty, witness protection and/or cash reward. A serious reward, in the millions. Such a rule would make it harder to pick henchmen, since in effect you're making them a millionaire if they turn on you.
Indeed, you have to worry that the very reason they might be joining you so easily is they plan to rat you out for the millions.
We have had a program for whistleblowers on companies that try to cheat the government, with some success, but I might suggest a good place to try a henchman law would be vote fraud. Combined with very stiff penalties (including life in prison for major vote fraud, which is effectively a coup d'etat) this should keep the conspiracies small, and less powerful.
Howoever, hacking of voting machines at the factory is a special type of vote fraud which could be done by a very few conspirators for a big result, so we still need voter verified ballots.
Of late, I've become curious about how readily we could, using certain encryption techniques, design a ballot system which would let you change or repudiate an already cast ballot, but still preserves the important principles of secret ballot. As you might guess, that's a challenge, since it seems inherent that if you can reverse a ballot you can know what it is. But it's possible some blinding algorithms might allow this.
Another option for mail-in ballots would be to provide each mail-in voter with two or more mail-in ballots. The ballots would be marked with a code, such as a red or green sticker. The voter would be told, through a secure and in-person channel, that one of the ballots is real and the others false. They must remember the colour of the true ballot. If they send in the ballot of the wrong colour, it also contains an encoded number that only the counting office is able to tell tags it as a bad ballot (and a sign of attempted voter coercion.)
The vote-coercer might have you send in all your ballots. One defence would be to have this void your vote. This allows vote coercers to use force or money to stop people from voting (which they already can do) but not to vote a particular way. Combined with my plan for secure internet voting it could allow one half of the secret ballot equation with votes cast over the internet.
Should we be able to perfect this, by finding a means for an audit trail on internet voting, this would, aside from saving much money, eliminate the problems of lines at the polls.
Submitted by brad on Thu, 2004-11-04 15:31.
We see the talk of an America divided in 2, but in fact it's not. There are more viewpoints than that. Normally a 2 party system tends towards the middle, this election was unusual in having a larger than normal difference among the candidates.
But perhaps now is the time to take the Democratic energy and try to push it into a movement for real reform. Not ballot recounts, not crazy dreams that can never happen.
By that, I mean getting at least one state to move to a preferential ballot system, such as Australian "Instant Runoff," Approval or Cordorcet, with an unfortuantely complex additional rule for how to cast in the electoral college when done.
Reforming the electoral college is unlikely (though an interesting hack is discussed elsewhere in this blog). 3/4ths of states must ratify any change to the college, and the small states would need a big constitutional price in exchange for stripping themselves of the extra power they have in the college.
However, individual states can change how they select their electors through ballot resolution or legislative action. Entirely on the local level. Ballot resolution seems the simplest approach. The only thing standing in the way is that many voters get confused by instant runoff systems. Basic Condorcet is easy to understand, but the tiebreaker modifications are often hard to understand. Still, the Australians manage it.
The first effort will probably fail and only educate the public. Eventually, some state, probably a small one, will go over and have such a ballot. This in turn will start to educate the rest of the nation. The ideas, once understood, are good ideas, and will appeal to the populace. It's hard to argue against them.
However, the 2 major parties will _want_ to argue against them because they are bad for those parties. In many elections, there is somebody who won because there wasn't a preferential ballot. In particular, Bush in 2000 and (arguably) Clinton in 1992. (On the other hand, Bush the Elder arguably _lost_ because there was such a system, and thus might support it.)
That's why a ballot proposition is the right way to do this. read more »
Submitted by brad on Tue, 2004-11-02 07:29.
Idea futures are interesting, so I went to tradesports.com to buy Kerry Futures contracts, since I believe pre-election polling is notoriously poor in quality. They were trading at 43 for a contract that pays 100 if he wins, and that seemed a good buy. By the time I could buy them (slow site, overloaded) they were 50 but now they are at 71. The bidders clearly have developed a strong reversal of feeling.
One source is the exit polls. While U.S. media are not publishing them, the CBC in Canada has few viewers in the USA and has no reason not to, and the news is good for the Senator, particularly in Pennsylvania.
Many have argued that idea futures with real money on the line are very good predictors. However people also clearly follow the polls, which is why Bush led in those futures for most of the election race, and even up to just a few hours ago. Still, Bush went south in these long before other sources so they're a good source as far as that is concerned. And perhaps I made a few bucks. (Not a lot, they limit you to $250 credit card if you do instant signup.)
Submitted by brad on Mon, 2004-11-01 08:25.
Today the word went around about mypollingplace.com, a site that helps people find their local polling station, running out of bandwidth from their provider and needing somebody to help it.
At the same time I did an interview for opinions about bitTorrent, and the idea came to me that a really useful application would be a P2P generalized web hosting tool aimed at spikes.
Volunteers who have a reachable IP address (including people who can open holes in NATs) would install software to volunteer to help host sites in need during a spike. This could be client software in a browser or permanent server software.
Then a few different things could happen. Ideally a site would install a tool which activated the sharing network when their load got high (but well before their provider cut them off.) People fetching static pages that weren’t sharing bandwidth would be redirected to a port willing to share. The page would also be modified to encourage folks to join the process. Folks who were willing to share might get access or be redirected, and they would cache the page, declare where it resides and serve it there for some period of time or number of hits.
If a site didn’t install the fancy redirector, they could redirect all web hits to a volunteer tracker that would do a further redirect to a real location willing to host the item.
Finally, for sites totally unaware of this, a browser plug-in could notice when access to a page is very slow or gets a “bandwidth limit exceeded” message. It could then query to see if any of the P2P folks have cached the page, and fetch from there, or offer the user a button to check or fetch from that network.
This latter mode is good because it strongly encourages having the magic plug-in with your browser. If you don’t have it, you can’t get at the P2P cache.
I’ve been informed that Coral from NYU does some of this, though it doesn’t do the “on demand” aspect that leaves the web in normal state when not overloaded, and switches to cacheing only when needed. With Coral and other redirects, your Google adsense doesn’t show up either! read more »
Submitted by brad on Mon, 2004-11-01 03:19.
There has been much writing (including here) about problems with the Electoral College in the USA, and I've even proposed solutions such as a tiebreaking system for close votes. I also noted the amazing coincidence that in the 4 times the winner of the college lost the popular vote, 3 were the 3 times we had a son or granson of a President elected.
But I thought it might be worth exploring the merits of the college, even though most individuals want it abolished. (Though no smaller states want it abolished since it gives them disproportionate power.)
First of all, there is the "official" goal of using the college system. It requires the President, who must win a true majority of the college, to be popular in at least half, and probably more, of the country. The framers didn't want it to be possible for a candidate with extremely strong support in one particular region to win the Presidency.
Example: Say a candidate, coming from a particular region, had immense support in that region, getting 90% of the popular vote, and much lower support (10-25%) outside that region. Such a candidate could win the popular vote since all those votes in their own region would count, even though they are not a national candidate. To win the college, their "region" would have to contain half
the population of the USA.
(This is based on the traditional, but not required all-or-nothing allocation method, which states do because it makes candidates want very much to please them.)
I'm not a fan of some regions having more power than others because of the political legacy of how big states were. 2 senators for Wyoming and 2 for California is grossly unfair. However, the concept of federalism does require some protection for regions, so you don't get one region ruling the land at the expense of another, which can lead to seperatism.
Another benefit is election cost. Due to the college, candidates know to spend their election money only in undecided, "swing" states. As such, they can campaign with much less money. With a popular vote, or even a proportional electoral college, they would have to campaign everywhere. This of course has downsides in terms of fairness, but if they had to campaign everywhere they would have to raise even more money, and be more beholden to the special interests that gave it.
Submitted by brad on Sun, 2004-10-31 15:45.
Tonight, I saw for the first time (for me) a drive-by trick-or-treating. I'm not talking about the growing phenomenon where parents drive their kids to wealthier neighbourhoods for a better class of candy. We had put out a ghost made from gauzy material with a very bright cold-cathode light inside, and hung it over the street. As I stood on the street a minivan pulled up and quickly stopped. Two children went to our front door and Kathryn gave them candy. Then we watch them get back into the van and it continued down the street, out of sight. The appeared to be cruising and stopping at houses with decorations they noticed, which can be found in many neighbourhoods. read more »
Submitted by brad on Sun, 2004-10-31 08:36.
You may have heard of the vote-trading concept, where a voter in a "decided" state (whose vote will make no difference) who wishes to vote for a major candidate pairs up with a voter in a swing state who wants to vote for a minor candidate. The idea is they swap choices. The swing state minor party supporter votes for the major party candidate (typically Kerry this time) and has a chance at making a difference in the swing state. The decided state voter, facing a large likely margin in their own state, votes for the minor candidate, boosting the national popular vote total for that candidate, which is all minor party voters care about since they don't plan to win.
Sounds good, but as you will see at votepair.org, they have 19,700 waiting and only 2300 pairs matched up, so the vast bulk of eager Kerry supporters won't get to swap.
There are many reasons to explain this. Some are simple. There are many more Kerry supporters than minor party voters. There are many more people in decided states than in swing states. And of course some people may wonder how much to trust the honour system used in vote swapping. You are supposed to meet your counterpart, and judge that they aren't a republican trying to play tricks, though in theory there is not much for them to gain by doing so.
One problem is the trade offers a lot to the Kerry supporter -- trading away a meaningless vote that can't change the results for a precious vote in a swing state -- but offers effectively nothing to the minority voter, since it doesn't affect the vote total for the minority party. (Indeed, it may hurt the minority party's prospects in that state.) As such, you are only going to get people who were more than happy to vote Kerry anyway, as they now will feel they gained from the trade.
Of course, a real market in votes is illegal, as these orgs know, but you can't ignore the realities of what makes a market work.
As such, the sites should consider offering a larger trade. They should let minority voters set a price for their switch, which could be 1.5, 2 or even 3 minority votes in the decided states. Then both sides would gain and it could up the count. (Libertarians, who value deeply the concept of win-win contracts as a basis of society, would be particularly swayed by this, and they are one of the largest minority parties.)
The existing 2300 might still be willing to do it for just one vote (remember, they probably were considering doing it for nothing.) Others could ask higher and higher "prices." Of course ask too high and you won't get anything, the voters will be allocated first to those asking less.
Yes, you can do a 1 for 1.5 swap, and I think that's the easiest and fairest. In this case each minority supporter is given two people to arrange swap with, one fully, and the other 50%, in that this other is expected to have an identical 50% deal with another supporter of the same minority candidate.
(Ethically, after you arrange things with the first voter in a 50% partnership, you would need to tell them if the 2nd voter fell through, and get it resassigned. Ditto for a 2 for 1 exchange.)
Submitted by brad on Thu, 2004-10-28 15:28.
We've seen the explosion in Voice over IP phone companies, using lower IP costs (and regulatory bypass) to offer very cheap long distance. Today, in the wholesale market, you can place VoIP calls to regular phone numbers in the major cities of the world for between 0.5 cents and 1 cent per minute. So cheap that companies are routinely offering people "unlimited" long distance plans for a flat monthly fee.
The rates are cheap, but they aren't yet free, so calls don't happen without contracts and plans and arrangements.
Here's something begging to be done: The cellular carriers in the USA and Canada should allow people calling in from the internet to call any cell phone for free.
This would cost them the tiny amount they get for terminating long distance calls to cell phones, but the end result would be their own customers billing more minutes. That's what they want (sort of.) They could release a free program for use on common PCs to call any cell phone. They could have a java applet or ActiveX control to make a web page to let you call any cell phone for free, just as they let you send an SMS from a web page for free. (If you have a headset with a mic, at least.)
But allowing free and open calling would encourage lots of innovating applications from the marketplaces. Smart PBXs would coordinate and connect with company cell phones over the internet. More advanced apps would link cell phones closer to PBX functions. People who use their cell as their home phone would have another reason to do it -- now all their friends can call free from VoIP phones or any PC. Companies like Vonage would offer free calls to cell phones even for people not on the unlimited plan. read more »
Submitted by brad on Thu, 2004-10-28 13:23.
Soon as it was out, I bought the EOS D20. I sold my D60, which I had replaced my D30 with, so I am obviously generally pleased with Canon's line. The new camera has a lot over the D60 -- 2 more megapixels (or 3500 high for panoramas), much better low-light shooting ability with low-noise high-ISO, and fast shooting (5 frames/second for 25 frames.) It also has better focus, better controls, and an orientation sensor, something I've been wanting for a long time.
The orientation sensor is botched though. The menu offers the ability to rotate pictures in the display based on the orientation. Not what I want, so I turned it off. Sadly, this ended up turning off even the recording of the orientation sensor, which is stupid. You want this screen rotate off because it makes the picture half the size to do it, and it's wrong if the camera is being held in portrait mode, as it may be on a tripod. The right thing to do would be to rotate based on the camera's orientation -- if I shot in portrait and I show in portrait it should fill the screen. However, generally I don't want the display rotate on the small screen, I can just turn the camera in my hands.
Turning off the sensor recording without this mode is incredibly stupid, I hope they fix that in future firmware. Frankly, I would have them rotate the actual picture as they store it with a lossless rotate. It's a pain to have to run special software (theirs sucks) to convert all the pictures after you shoot them. Part of why I wanted the software was to get away from this annoying task.
For a decade or more, video cameras have come with a small pinhole next to the screw tripod mount. Mounting plates have a pin that goes in there to mount the camera securely. Why do still cameras never come with this hole? It would make the mounting much more secure.
The camera has a full zoom in, something the Canon DSLRs were very slow to adopt. But nobody yet has a "smart" zoom-in, which zooms directly to a 1:1 view of the highest contrast element in the scene, or alternately to one of the focus points that it did the autofocus on (though that can move.) This would let me quickly see if shots were blurry or not. If the highest contrast portion of the scene is not in focus, the whole thing is blurry. If it's in focus but was not at the distance of my subject, I know I shot the wrong thing, and can correct it now. You don't see that on the small screen.
Can't say I'm thrilled with combining the on/off for the back-wheel with the on-off for the camera. I always want that wheel on.
And yes, it's time for DSLRs to get live preview. They can do it, they already split the beam a couple of ways -- most of the light to the eyepiece, but part of it to the autofocus/light meter instruments. Low-res sensors are now cheap enough that you could put one in to perform all those roles and also allow live preview. Live preview isn't just for cheap point and shoot cameras. Yes, through-the-lens is better for composing most shots, but I love how on my G5 I can shoot in places I can't get my eye to the viewfinder, like when the camera is held above my head, at my waist, on the ground, on my knee for stability etc.
And nobody has yet imported features like Nikon's Best-Shot-Selector, a bracketing mode which picks the least blurry of a series of shots in low light. Of course you can do this by just keeping all the shots.
But as I said, I am generally satisfied, but it doesn't mean they can't keep improving.
Submitted by brad on Mon, 2004-10-25 17:08.
Many people like the idea of soda fountains, but the official fountains that mix soda and pre-mix concentrate are quite expensive and work to maintain.
To gain some of the convenience and efficiency, how about a quasi-real fountain using existing 2-liter pop bottles? Such a fountain would be designed to look like a regular one. Inside, ordinary 2-liter bottles would be attached. In a "party" model, they would just be in an icebox, in a full-time model it would have refrigeration.
To attach the bottles, their caps would be removed and a special cap screwed on that would have two hoses -- one for pop to come out, and the other for gas (CO2 or air) to come in to push out the drink. A CO2 cylinder could power this, or an air pump if you want battery power. Gravity might be able to do it but pressure certainly could.
The special caps would have another quick connect interface that lets them be turned upside down and snapped into place. Alternately, the bottles could be in another orientation as long as the feed tube can reach their bottom.
You might want to be able to have two bottles of the same drink feed the same fountain to give more capacity and the ability to drain one and then the other so that they can be "hot swapped" (really cold but you know what I mean.) Or people might prefer to be able to offer lots of different drinks, but just 2L of each.
The goal of this design is to be very cheap. The one based on ice and CO2 cartridges would be very cheap and need no power -- more of a party amusement, but also an efficient way to give people only what they need.
Of course you could just empty the bottles into reservoirs, but then you need a way to clean them. Cleaning the tubes could be done with an attachment for a garden hose to flush them, and the caps could go in a dishwasher.
This is my lowest tech idea yet...