Tuesday, November 27, 2012

Internal Division Errors

A while ago I started a project to redraw the counties and county-level equivalent areas in each of the 50 states.  I have several reasons for undertaking such an odd task.  First, I like maps and playing with maps.  Second, it is way for me to virtually visit places I'll never have a chance to get to in real life.  Between the various online map layers, one can get a decent sense of what just about any area of the country is like.  Third, in many states, reconfiguring the internal divisions is something that should be done.  Some counties are too large, others are divided by nearly impassible geography, many could be re-drawn with more natural boundaries, and quite a few are very under-populated.  The new boundaries attempt to balance size, population, watershed boundaries, economic interconnections, and other factors.  Not every newly-drawn or re-drawn county is optimal, but I feel the states I have completed are much better divided than than they are now.  Eventually I'll get to all of the states, but it will take a few more years.

Due to limitations in GMaps, some states may be split into two pages.  In the case of California, I manually split the map up to avoid having to jump around between three or more pages.  You can view more than one of the maps at a time by bookmarking each one in your GMaps account, and then clicking on each in your "My Places" list.  It's kludgy, but it works.*

  • Alabama counties
  • Alaska counties
  • Arizona counties
  • Arkansas counties
  • California counties (Northern, Central, Southern)
  • Colorado counties
  • Connecticut counties
  • Delaware counties
  • Florida counties
  • Georgia counties
  • Hawaii counties
  • Idaho counties
  • Illinois counties
  • Indiana counties
  • Iowa counties
  • Kansas counties
  • Kentucky counties
  • Louisiana counties
  • Maine counties
  • Maryland counties
  • Massachusetts counties
  • Michigan counties
  • Minnesota counties
  • Mississippi counties
  • Missouri counties
* - If you have a GIS program installed, download the KML for each state, and then load them all into a new project, along with state and county layers.

Monday, November 19, 2012

Picture This and That: November 2012 Comparison Chart Edition

Note: I post updated versions of this chart every six months or so.  Check the photography category for the latest.

Photokina has come and gone for the biennium, and we're now entering the season of the holiday consumption orgy.  To assist gift-givers and gift-demanders, below is a chart of the available (or announced and shipping before Christmas) interchangeable-lens cameras (ILC).  There are a slew of newcomers this time, and I have added a row for entry-level full-frame digital single-lens reflex (DSLR) cameras to accommodate what seems to be a new market category.  I also added a row to the rangefinder-style mirrorless system camera (MSC) section because Sony filled a gap in their pricing structure. I've left in place the row for the now-empty full-frame flagship studio DSLR category as I anticipate Canon and Nikon will fill it again.  I am still excluding Fuji's X-series (I am so, so tired of "X" everything) even though a second, less expensive model was added because... well, because it would screw up my chart.  Maybe next time.

On to the newcomers, with the old models in parentheses.  Canon introduced the EOS-M (n/a), EOS 650D/T4i (600D/T3i), and EOS 6D (n/a).  Nikon introduced the J2 (J1), V2 (V1), D3200 (D3100), and D600 (n/a). Sony introduced the NEX-F3 (NEX-C3), NEX-5R (NEX-5N), NEX-6 (n/a), and SLT-A99 (n/a).  Pentax introduced the Q10 (Q), K-30 (n/a), and K-5 II (K-5).  Olympus introduced the E-PM2 (E-PM1) and E-PL5 (E-PL3).  Panasonic introduced the DMC-GF5 (DMC-GF3), DMC-G5 (DMC-G3), and DMC-GH3 (DMC-GH2).  And, finally, Samsung introduced the NX1000 (NX100), NX210 (NX200), and NX20 (NX10).  Canon discontinued the EOS 1100D/T3 and reduced the price on the EOS 600D/T3i to replace it.  There are no direct replacements for the Canon 1Ds Mk.III, Nikon D3x, Olympus E-P3, Olympus E-620, and Pentax K-r. As always, older models are often still available new, and (initially) at a discount relative to current models.  (Old new stock tends to get more expensive several years out.)

A previous version of this chart is here, and a general background post can be found here.  If you're a first-time ILC buyer, remember that cameras are just tools, and the raw pixel count is very likely not the most important selection criteria.  Along with selecting a body, it's also important to consider what gets attached to the front of the camera: glass.  And one shouldn't forget to budget for accessories when making a decision. A minimal accessory package should include two storage cards, one spare battery, a UV filter (a.k.a. scratch protector) for the lens, and a basic carrying case.


Big 2UpstartLittle 4
Camera categoryBrand
(2010 ILC
market share)
Canon
(45%)
Nikon
(30%)
Sony 3
(12%)
Pentax
(?)
Olympus
(5%)
Panasonic
(?)
Samsung
(?)
RF-style MSC
price w/ zoom 1
Low-end-J2
10.1MP
$550
NEX-F3
16.2MP
$500
-E-PM2
16.1MP
$600
DMC-GX1
16.1MP
$470
-
Mid-Range--NEX-5R
16.1MP
$650
Q10
12.4MP
$600
E-PL5
16.1MP
$700
DMC-GF5
12.1MP
$520
NX1000
20.3MP
$600
High-endEOS-M
18.0MP
$799 4
V2
10.1MP
$800
NEX-6
16.1MP
$1000
K-01
16.3MP
$830 2
--NX210
20.3MP
$800
Flagship --NEX-7
24.1MP
$1300
----
Consumer
DSLR / SLT /
SLR-style MSC

price w/ zoom 1
Beginner600D/T3i
18.0MP
$500
D3200
24.2MP
$650
SLT-A37
16.2MP
$600
----
Mid-Range650D/T4i
18.0MP
$800
D5100
16.2MP
$650
SLT-A57
16.2MP
$700
K-30
16.3MP
$850
E-M5
16.1MP
$1100
DMC-G5
15.8MP
$800
-
Enthusiast60D
18.0MP
$1200
D7000
16.2MP
$1300
SLT-A65
24.3MP
$900
--DMC-GH3
16.1MP
$900
NX20
20.3MP
$1000
Professional
DSLR / SLT

price body only
Mid-size
crop sensor
7D
18.0MP
$1600
D300s
12.3MP
$1700
SLT-A77
24.3MP
$1300
K-5 II
16.3MP
$1200 5
E-5
12.3MP
$1700
--
Entry-level
full frame
6D
20.3MP
$2100
D600
24.3MP
$2100
-----
Mid-size
full frame
5D Mk.III
22.3MP
$3500
D800
36.3MP
$3000 5
SLT-A99
24.3MP
$2800
----
Flagship action1D X
18.1MP
$6800
D4
16.2MP
$6000
-----
Flagship studio-------

1 - If multiple zoom kits are available, the price is for the cheapest, which usually includes an 18-55mm (equivalent) lens.
2 - dual-lens kit
3 - Sony's web site is horrible so I'm not linking to it.
4 - prime not zoom lens
5 - The K-5 II and D800 are available without an anti-aliasing filter for a modest additional price.

Wednesday, November 7, 2012

Glad to Be Wrong: 2012 Election Edition

As something of a pessimist when it comes to predictions, I'm always glad to be wrong.  Last week, I predicted a narrow win for Obama.  But the President won the Electoral College handily, 332-206, verses my prediction of 290-248.  And I was two seats too negative about the Senate, which I predicted to be split 53-47.  I was about right with the popular vote, which Obama won 50.3% to 48.1% to 2.6% (mainly for Gary Johnson), verses my prediction of 50% to 49% to 1%.

Unfortunately, I was several seats too positive about the House, as the Democrats only picked up seven seats in the House of Representatives.  I predicted a 230-208 split, but the final result looks like it will be 235 to 200 (barring any flips during a recount, which is unlikely).  The Republicans appear to have done a very good job of protecting their candidates during the 2011-2012 redistricting process.  This will make actually governing the country even harder than I expected, because the remaining Republicans will have even less reason to fear the displeasure of their constituents.  Another reason Republicans won't worry about reelection in 2014 is that there is a historical pattern of the president's party losing House seats in mid-term elections.  I haven't looked at the details yet, but I am fairly sure that none of the Republican freshman are moderates, so the caucus will be even more radical.  All this points to gridlock no matter how many generous offers of compromise Obama makes.

I'll be happy to be wrong about my prediction of gridlock, though it will take a bit longer than a week to see the outcome.

Tuesday, November 6, 2012

Election of Perfection

While the 2012 election isn't the first in America where the mechanics of voting make for news stories, this is the year when I've finally had enough of the doubt-inducing hiccups and quirks.  We really need to move to a uniform national system based on paper ballots.  Given America's overall election process, here's how I think it should be done.

Prior to election day:
  1. States run their nominating processes, and finalize the ballot seven weeks before election day.
  2. Ballots are printed and verified.
  3. Early voting runs from from five weeks until one week before election day.  Ballots are placed in boxes, which are secured by well-known procedures, by the voters.
  4. Absentee ballots are issued starting five weeks and ending one week before election day. Ballots are accepted unconditionally until close of voting on election day, and provisionally based on postmark for a week after election day.
  5. Optical scan machines are configured, tested, and secured by one week before election day.1
  6. Applications by groups wanting access to the paper ballots during the four weeks after the election are accepted until one week before election day, and issued by the day before the election.2
  7. Voter rolls for polling places are generated.
  8. Scan machines are set up the day before election.
On election day:
  1. Voting begins.
  2. A small random group of precincts (1 to 3 percent) are selected for hand counting.3
  3. Early and absentee ballots are scanned.
  4. Voting ends.
  5. Vote totals are read off the scan machines and reported, except where hand counting is being done.3 
  6. If a machine fails, or if the number of write-in votes exceeds the votes for each of the other candidates, the precinct is counted by hand.
  7. The hand counts are reported.
  8. The hand counts are verified using the scan machines, if possible.
  9. The optical scan machines are inspected for tampering, and then moved to a central site or sites.
In the week after the election:
  1. Any failed scan machines are replaced with standby machines.
  2. A second random group of all precincts are hand-counted.
  3. The other precincts are verified by running the ballots through the scan machines again.  If the number of write-in votes exceeds the votes for each of the other candidates, the precinct is counted by hand.
  4. The optical scan machines are re-inspected for tampering.
In the next three weeks:
  1. If a recount is deemed necessary, a third random selection of precincts are counted by hand, and the rest are re-verified using the scan machines.
  2. Write-in votes are tabulated.4
  3. Challenges are considered for provisional in-person and absentee ballots.
  4. Paper ballots and optical scan machines are made available in controlled conditions for any pre-qualified group to examine.2
  5. Provisional in-person and absentee ballots are counted two business days before final results are issued.
  6. Provisional ballots are recounted one day before final results are issued.
  7. Final results are issued four weeks after the election.
And finally:
  1. Scan machines are released for repair and reprogramming.
  2. Paper ballots are archived for 10 years, but can be viewed under controlled conditions by any member of the public.
The process outlined above is off the top of my head.  If I have enough energy, I will compare it to the procedures devised by professional voting integrity groups.

1 - Machines of any sort are in no way necessary for successfully conducting an election.  But Americans are impatient and like technological "fixes", so I've included optical scan machines.
2 - The point of permits is to prevent chaos during the post-election period (which may include a recount) while still allowing public access.
3 - A small group of precincts should be hand-counted on election night so that any gross discrepancies between hand and machine counts are detected immediately.  It should be done before the totals are read off of the machine so that vote counters have no expectations of what the outcome might be.
4 - In almost every election, identifying and tabulating the write-in votes is a meaningless exercise, but for whatever reason it is done.

Updated 2012-11-08: Added more steps.
Updated 2012-11-12: Added more steps.

Thursday, November 1, 2012

Odds of Sods: November 2012 Edition

In the three preceding post I have made my predictions for the outcomes of the contests for the Presidency, the Senate, and the House of Representatives.  Below is how I see the cycle as a whole.  The second best outcome, which preserves the status quo, is also the most likely.  The very best outcome - full Democratic control - is the third most likely.  Unfortunately, the sixth best (or third worst) outcome is the second most likely.  The chances of the other outcomes are very low at this point, as it would mean that people predicted two or more of the three contests wrong.  Modern polling is reasonably good, and there are enough of them for the Presidency and Senate that the degree of confidence for individual contests is fairly high.  The House is a little more dicey, as there aren't as many polls for individual races.  In fact, most professional poll crunchers avoid the House altogether, and others look just at the overall picture.  The redistricting process creates some additional uncertainty in the meaning of the generic congressional ballot, but I still think it indicates moderate gains for the Democrats, if not as many as I'd like.  But, overall, I think Democrats should be moderately pleased at the end of next Tuesday night.


OutcomeLegislative OutlookImplication for CountyOdds of Happening
DemocratsRepublicans(1 = best/highest, 8 = lowest/worst)
President, Senate, House-Slow advance13
President, SenateHouseNeutral21
President, HouseSenateNeutral36
PresidentSenate, HouseSlow retreat44
Senate, HousePresidentFast retreat57
SenatePresident, HouseFast retreat62
HousePresident, SenateFast retreat78
-President, Senate, HouseThe Apocalypse85

One Office to Herd Them All: November 2012 Edition

As is the case every four years, the grand prize for the election cycle is the office of the President of the United States of America.  This cycle pits the Democratic incumbent, Barack Obama, against the Republican challenger, Willard Romney.  The key to understanding the presidential race is that it is 51 state races, not one national race.  Given the electoral college system, which I really don't like, I believe (as of now, November 1st) that Obama will be the winner on November 6, with 290 electoral votes to 248 for Romney.  The best outcome Obama is likely to get is 332 to 206, and the worst is 265 to 273.  270 EVs is the threshold for winning.  269 EVs for each would send the election to the House of Representatives and the end of the world as we know it.  (I'm kidding about that last bit.  I hope.)  In reality, the chances of Obama losing in Nevada are less than 15%, so it would have to be a very bad night for the Democrats if Obama were to receive less than 271 EVs.  A lot of current polling gives Obama a victory in Virginia, but I think he will just barely lose.  Winning that state would put his EV total over 300, which doesn't mean much by itself, but people tend to fixate on nice, round numbers.

I'm not going to analyze the popular vote because it's complicated and it doesn't matter, strictly speaking.  It does matter a lot for legitimacy purposes, but only for Obama.  My guess is that the result will be about 50-49-1% in favor of the incumbent, which will make him a legitimate victor in the minds of Democrats and true swing voters, and would mostly shut the media up.  If Obama loses the popular vote but wins the electoral college, the press will insist that he bow to every Republican whim, which of course they did not do for Bush the Lesser.  That would be both entirely unfair and entirely unsurprising, because media has two separate standards for the two parties.

On election night, the first state to watch for will be Virginia, where polls close at 7PM EST.  If Obama takes that state, he'll almost certainly have won the election (given the near-certainty of Obama winning the 265 EVS in the higher confidence categories).  The networks may also call Florida at 7PM EST, but most likely they will wait until 8PM EST when polls in the panhandle close.  Florida is another state that would basically guarantee that Obama was victorious.  If Obama loses in both of those states, he will need to win Colorado, Iowa, or Nevada.  Those states close their polls at 9PM EST, 10PM EST, and 10PM EST respectively.  New Hampshire polls will close at 8PM EST, but by itself the state's 4 EVs cannot give Obama the win when added to his baseline of 265.  I think we'll have to wait until after 9 PM to see the good guy win.

 Certain Dem.  Likely Dem.  Lean Dem.  Lean Rep.  Likely Rep.  Certain Rep. 
Alabama9
Alaska3
Arizona11
Arkansas6
California55
Colorado9
Connecticut7
Delaware3
District of Columbia3
Florida29
Georgia16
Hawaii4
Idaho4
Illinois20
Indiana11
Iowa6
Kansas6
Kentucky8
Louisiana8
Maine *4
Maryland10
Massachusetts11
Michigan16
Minnesota10
Mississippi6
Missouri10
Montana3
Nebraska *5
Nevada6
New Hampshire4
New Jersey14
New Mexico5
New York29
North Carolina15
North Dakota3
Ohio18
Oklahoma7
Oregon7
Pennsylvania20
Rhode Island4
South Carolina9
South Dakota3
Tennessee11
Texas38
Utah6
Vermont3
Virginia13
Washington12
West Virginia5
Wisconsin10
Wyoming3
Subtotals21748254237169
Baseline total
265

206
Grand TotalObama
290248
Romney

* - By state law, both Maine and Nebraska can split their electoral votes, but that won't happen in 2012.

One Hundred Mini-Presidents: November 2012 Edition

As is the case every two years, approximately one-third of the Senate is up for re-election in 2012.  Below is what I think (as of now, November 1st) the outcome on November 6th will be.  Since the last version of this post, the outlooks for four contests have changed significantly. I've moved one Democratic-held seat (VA) from tossup flip to tossup hold. I've moved one Democratic-held seat (CT) from certain hold to tossup hold.  I've moved one Republican-held seat (IN) from Republican hold to Democratic flip.  And I've moved one Republican-held seat (AZ) from certain hold to tossup hold .  The result will be three D-to-R flips (MT, ND, NE) and three R-to-D flips (IN, MA, ME), with the final distribution for the next Congress 53-47 in favor of the Democrats (which includes the two Independents).  Previously I viewed VA as a flip and IN as a hold, resulting in a 51-49 partisan split.  I think the best outcome possible for the Democrats is 55-45 (winning in AZ, MT) and the worst is 50-50 (losing in CT, IN, VA).

The main reason for the changes is that the Republican candidates have been worse than I expected.  Flake, Mourdock, and Allen (plus Aiken in MO) are all terribly unlikable, so they've turned off independents.  On the other hand, McMahon in CT has done better than I expected, most likely because she's given herself huge amounts of money (over $42M so far).  I would like to see more of the western Democrats win (Carmona, Berkeley), but the polling doesn't show that they have much of a chance.   For two candidates where the polling is close (Heitkamp, Tester) I think their states are still too Republican for them to win on the same ticket as Obama.  For the other other three (Murphy, Kaine, Baldwin) their states are sufficiently purple or blue that they will win.  Donnelly, like McCaskill a few months ago, has been gifted a win because the Republican candidate said something vile about rape.

The Democratic caucus is likely to be more liberal after this election, with the additions of Murphy, Warren, and Baldwin, coupled with the losses of Lieberman, Conrad, Nelson, and Tester.  Hirono, Heinrich, and Kaine are likely to vote in roughly similar fashion to their predecessors, and King will reprise Lieberman's role as an annoyance that frequently provides a helpful vote.  Donnelly will be the sole conservative newcomer, though naturally he will be more liberal than the Republican he will replace. If Reid and the Democrats decide to ditch the filibuster (strong emphasis on "if") in 2013, at least judges and the like will be Democrats, and possibly even liberal Democrats on occasion.  The importance of those positions should not be underestimated.  Ditching the filibuster would also shift legislative control away from Conservadems (Pryor, Carper, Landrieu, McCaskill, Hagen, Manchin) who are numerous enough to give Republicans effective control on many issues if the filibuster remains.  But even if the rule is removed, the overall legislative environment would remain poor for most of the policies I'd like to see changed because the Republicans are likely to narrowly retain control of the House.


DemocraticRepublican
2011-2012 seats5347
Carryovers3037
Contested2310
Certain holdsCalifornia (Feinstein)
Delaware (Carper)
Hawaii (open D to Hirono)
Maryland (Cardin)
Michigan (Stabenow)
Minnesota (Klobuchar)
New Jersey (Menendez)
New York (Gillibrand)
Rhode Island (Whitehouse)
Vermont (Sanders*)
Washington (Cantwell)
West Virginia (Manchin)
Mississippi (Wicker)
Tennessee (Corker)
Texas (open R to Cruz)
Utah (Hatch)
Wyoming (Barasso)







Likely holdsFlorida (Nelson)
Pennsylvania (Casey)
New Mexico (open D to Heinrich)
Ohio (Brown)
Nevada (Heller)



TossupsConnecticut (open D to Murphy)
Missouri (McCaskill)
Wisconsin (open D to Baldwin)
Virginia (open D to Kaine)
Arizona (open R to Flake)
Montana (Tester (D) to Rehberg)


Likely flipsIndiana (R*** to Donnelly) 
Massachusetts (Brown (R) to Warren)
Maine (open R to King**)
Nebraska (open D to Fischer)
North Dakota (open D to Berg)
2012 results2310
Swing00
2013-2014 seats5347

* Sanders is an independent who caucuses with the Democrats.
** King is an independent who will probably caucus with the Democrats if there is a tie; otherwise he'll side with the winner so as to get a little more clout for his state.
*** Mourdock defeated Lugar in the Republican primary, and will lose the seat for Republicans.

Update 2012-11-02: Added sentence about Donnelly.
Update 2012-11-05: Corrected Baucus to Conrad.

435 Creepy Critters: November 2012 Edition

As is the case every two years, every member of the House of Representatives is up for re-election in 2012.  Below is what I think (as of now, November 1st) the outcome on November 6 will be.  I have been somewhat optimistic in my analysis, but even if it is accurate, the Democrats will fall short, leaving the Republicans with a slim 230-205 majority and an even more radical caucus.  That situation will not be good for legislation I think should be passed, as the Republicans will double down on obstructionism, hoping to duplicate the 2010 election in 2014.

I'm not going to break out individual races for the House because I just don't have the patience to document the 80 or so races that are somewhat competitive, or better.  At a higher level, the states to watch are Arizona, California, Colorado, Florida, Illinois, Iowa, Michigan, Nevada, New Hampshire, and New York.  Those 10 states have the most competitive races.  States definitely not to watch are Alabama, Alaska, Delaware, Hawaii, Idaho, Kansas, Louisiana, Mississippi, Missouri, Nebraska, New Mexico, Oklahoma, Oregon, South Carolina, South Dakota, Tennessee, Vermont, and Wyoming.  There no competitive races in those 18 states.  The remaining 22 have either only 1 competitive race, or races that are only somewhat competitive.

Because 2012 is a redistricting year, I have included a column for the change in the number of seats each state holds.  I have also broken the "swing" into two separate columns because, with redistricting, one party's gains in a state don't always equal the other party's losses.


2011-12
Democratic
Seats *
2011-12
Republican
Seats *
Change
in Seats
Comp.
Races
2013-14
Democratic
Seats
2013-14
Republican
Seats
Net
Democratic
Change
Net
Republican
Change
Alabama1616
Alaska0101
Arizona3513542-1
Arkansas1304-11
California341911371633
Colorado343431-1
Connecticut50150
Delaware1010
Florida6192610174-2
Georgia5811410-12
Hawaii2020
Idaho0202
Illinois811-161355-6
Indiana3627-11
Iowa32-1431-1
Kansas0404
Kentucky24124
Louisiana16-115-1
Maine20120
Maryland621711-1
Massachusetts100-1181-21
Michigan69-1368-1
Minnesota442531-1
Mississippi1313
Missouri36-126-1
Montana01101
Nebraska0303
Nevada1212312-1
New Hampshire022202-2
New Jersey76-1166-1
New Mexico21121
New York218-282251-3
North Carolina763310-44
North Dakota0101
Ohio513-2511
-2
Oklahoma1405-11
Oregon4141
Pennsylvania712-14612-1
Rhode Island20120
South Carolina151161
South Dakota0101
Tennessee2727
Texas92342122431
Utah121104-12
Vermont1010
Virginia38138
Washington5412641
West Virginia12112
Wisconsin35235

Wyoming0101
Totals193242020523012-12

* - I have added the five currently vacant seats to the party of the previous officeholder.