Retiring Dan Rather

As the paper this morning said, it will be 24 years since Walter Cronkite stepped down from the Achor char at CBS. I remember that day. I was quite young at the time, but I still liked Cronkite. The new guy proved himself, but never really won the affection Cronkite did. And sadly, it never really got a lot better then that.

The true low point for me was the 2000 presidential elections. As the wait for Florida grew into the night, Dan Rather got more and more southern, and weirder and weirder with his analogies. That was more of the stick that broke the camel's back. My parents had stopped watching Dan for some time and had turned to ABC.

The broadcast news networks do have some problems ahead of them, as anyone who has watched the half-hour newscast can tell you. Just looking at the ads in those newscasts tells you who their expected demographic is, the over 50 set. While that'll be good for another decade or two, it probably won't pay the bills for much longer. The future of News is already in flux thanks to my own generation's penchant for getting news online (and ironically enough, largely in text form).

Cars of the future

Slashdot had a story today about "Self-Navigating Vehicles". The article referenced goes into some details about cars that will drive themselves in the future. I've already covered some of this in my rant agains PRT. It covers some of the details about what needs to happen before we get there, and what progress we've already made.

One thing that really needs to happen is the ability to "drive by wire". Right now most stearing is still mechanical linkages assisted by a power-stearing unit. The same can be said for both breaks and accellerator. These are begining to change, especially in the newer hybrid vehicles, though it will still be some time before they are as common as, say, anti-lock breaks.

And I have to say that I'm all for such systems. The privacy concerns around the system will be a hard to work through, but I like the technology. Since I drive to work everyday instead of bussing (bussing from my new house takes 3x as long as driving due to transfers), this sort of thing is really appealing. I may yet get it before I retire.

Social security: defined benefit, or defined contribution?

One of Bush's major themes for Social Security is to 'create an ownership society'. To do that, social security funds would be invested directly into the market. This money would then be protected (i.e. no early withdrawal) until the worker reaches the appropriate status in life to receive their benefit.

This does recast social security to the currently misconcieved notion that money deducted from your paycheck now is what will pay your retirement (or disability, or...) later. Currently, that money is spent to pay current benefit receivers. It is the nature of the current system that creates the benefit gap in the future, as more and more money will be needed to support the boomers as they retire and the working-age population diminishes as a percentage of the total population.

The Bush plan is attempting to change the 'pay it forward' system we have now, into a 'pay myself later' system. On the surface of it, it sounds like a really nice idea. This way spikes (or dips) in birth-rate won't skewer the fund.

The big problem comes in converting the system. As retires and near-retires have noticed in the past 15 years, pension funds have been undergoing a revolution. Out are the defined-benefit plans of old, and in are the defined contribution plans such as the much vaunted 401(k). The defined-benefit plans of old took a drubbing in the recession of the early 1990's as the stock market and current contributions weren't enough to keep up with benefits being paid out. Then the stock market took off like a rocket in the mid '90s and some new savings instruments were created, such as the 401(k), 403(b), and the Roth IRA. Plans were converted, often without much warning, and the wailing and gnashing of teeth of those within a few years of retirement were heard.

Much can be learned from their pain. The big problem with defined-contribution plans is that in order to get a good benefit out of them, you have to participate in the plan for a couple of decades. The defined-benefit plans our parents and grandparents enjoyed paid out with as 'little' as 10 years of employment. The ways of handling this are varied, but came in two flavors. First, a line was drawn in which anyone hired after a certain date is in the new defined-contribution plan and people hired before that date still contribute to the old defined-benefit plan. Second, if the pension fund was solvent enough to handle it, the pension fund itself was restructured so that past contributions were allocated to the employee in a protected state until their retirement.

Social Security lacks something the old defined-benefit plans had, and that's the ability to invest towards the future. By law right now that investment is limited to US treasury notes, which while safe, underperform the stock market by a wide margin. Pension funds do not have that limitation, which has led to fund failures and bailouts in the past, as well as greatly improved solvency. This additional solvency means that current participants don't have to pay as much in order to match paid-out benefit. Since Social Security lacks this, restructuring it will be devilishly hard. If a line in the sand is drawn in which new contributors are shunted into a defined-contribution version of Social Security, it will be the continuously shrinking working-pool that has to 'feed' the continuously increasing retired-pool of workers. The fund is patently not solvent enough for a restructuring. And factor into all of this that AARP is a very powerful lobby in Washington, and you have a Prime grade mess.

So how the heck is this fund going to get converted? A bit at a time, and with some painful public funds kicked in to keep it running, that's how. In the beginning, the contribution to the defined-contribution portion will be a pittance as compared to the rest of it; at my salary I'll probably sock away a few hundred a year even with the employer-match. As time moves on and retires kick this mortal coil, the percentage of payroll deduction going into the savings fund will increase. This increase will, due to actuarial needs, be devilishly slow and measurable in multiples of decades.

So is this a good idea? Hard to say. The defined-benefit plans had a weakness in that if they were badly managed, the entire plan fails and everyone is out in the soup-line. The defined-contribution plans are weak in that if a worker doesn't work long enough, or picks funds poorly, they are individually out in the soup-line but the plan survives. Social Security as it sits now is facing the first problem since the ratio of retired-to-workers will be undergoing sea-change in the not too distant future, and that will force either greatly increased social security deductions (i.e. new 'taxes'), or greatly reduced benefits (i.e. AARP gets pissed). Rock, meet hard place. Clearly, Something has to be done.

Social Security was created in the shadow of the Great Depression. Because of this, trust in the stock market to be a reliable vehicle for long term savings was at a relative low. This caused, in my opinion, the rule that Social Security funds shall not be invested in anything as risky as stocks. Social Security was devised as a guarantee of an income in your later years. A very risk-adverse guarantee at that. As anyone who has had need to know the difference between 'Income', 'Value,' and 'Growth' classes of securities can tell you, 'income' offers highest short-term stability but at a lower long-term return, and 'growth' offers high long-term return but with greatly increase short-term volatility. Funds that Social Security can invest in now are on the very conservative end of the 'income' bracket. Which is as it should be, as it provides that guarantee.

The Bush plan introduces more individual risk into the plan as a hedge against avoiding a hideously expensive bailout in the future. But more risk is something that AARP and the Unions won't take well. The camel's nose under the tent-flap is the only way to get this plan off the ground without royally screwing current and soon-to-be-receiving recipients. Even with control of both halves of congress and the whitehouse, this will not be an easy sell.

Electronic voting -- bad idea

My day-job includes making sure digitally stored data stays stored and unaltered. We accept a certain loss-rate as a hazard of doing business. Because of this, I'm fairly aware of all the myriad ways that data can go screwy in electronic systems.

I can see very little use for pure-electronic voting, and this is why I think this.

Voting in this country is an anonymous (mostly) thing. You go into your polling place, get a ballot, make your choices, leave your polling place. All the voting judges know is that you arrived and were given one (1) ballot, and at the end of the day the total number of ballots cast matches that of ballots handed out (minus spoiled ballots). The only record of your selections for your ballot are recorded anonymously but permanently in that polling place and then aggregated and submitted on up the chain.

In addition to the above, the recorded vote must be in a format that can be re-counted later should some provision of either the election process or law require recounting.

There are many, many ways that electronic data can go missing or change suddenly in such a way as to invalidate the above. Everything from malicious programmers sneaking in code to shift votes one direction, to insiders changing elections data outright once it is aggregated, to maintaining the technology can cause fault. The first, primary, foremost problem is that electronic data is very hard to make permanent. The second main problem is that electronic data is very easy to change once you have the right level of access.

It is a matter of course that the more complex a system gets, the more interesting ways it can fail. Prior to electronic voting, the most complex voting method was the mechanical lever system; which was prone to faulting and the prime reason it was phased out in favor of more efficient methods like optical-scan. Unfortunately, mechanical-lever is presumed to be the most accurate of voting methods due to the greatly reduced number of 'spoiled' ballots it produces.

Physical-methods of voting such as punch-card, optical-scan, lever (which produces a machine-readable tape, dials on the back, or something similar), and hand-counting, rely upon the counting of physical somethings and aggregating that number to arrive at a winner. Recounts are just that, the physical somethings are looked at again and the numbers jiggered as needed. The ballots or voting records are preserved for the express purpose of validating the election.

On the surface of it, electronic voting allows for great improvements in how elections are processed. As with lever, voters hit the button for their candidate and their vote is recorded. With enough technology, that vote can be transmitted in real time to the relevant voting authorities and therefore does away with the hand-count that happens in polling places after close. With zero spoiled ballots, the need for recounts is greatly reduced, and in the case of a recount can be performed in a matter of hours.

The problem is the permanent record of the vote. The most widely recognized method of making data permanent is to burn it to CD. In the physical schemes the permanent record is recorded at the time of voting. In the electronic methods this is a lot harder to accomplish. At the end of the voting period each electronic voting station may spit out a CD with that station's votes on it, akin to the lever-station's paper-tape.

Also, you have to trust, completely, that the software and hardware handling the voting process is 100 percent trustworthy. If you can't trust the voting machine to record your vote correctly, then the process is flawed. This is why the software and hardware that run electronic polling stations needs to be open-source, so it can be peer-reviewed, and act as a verification as to whats running on the actual stations on election-day.

In my opinion, the 'best' permanent record of the vote would be a voter's receipt. Think check-out counter printer, but with your votes clearly marked on it. Should that receipt not match what you voted for, you get the chance to do it again. Once you've verified your receipt matches your votes, the receipt goes into the recount-box. At the end of the day your station reports the vote totals electronically, and then are further electronically reported up the chain. Should a recount be called for, or an audit of the effectiveness of the electronic system, the physical receipts can be hauled out for a manual recount.

This is because that despite all of our wonderful technology, paper really is the best archival mechanism for long term storage. For partially anonymous yet absolutely critical systems such as elections, it is the most reliable method next to rock-carving. Electronic data is fundamentally mutable, and elections data has to be fundamentally permanent.

Any electronic system really should be just a shell around the creation of a physical record. In the end, that physical record may be a bit of digitally-recorded/machine-readable (note: digitally recorded, not electronic. Barcodes are digital) something, or a bit of paper with the voter's preferences lined out. Once all that is done, it is a lot easier to trust the electronic system into the permanent physical record.

Unfortunately, there is another barrier to this. For the same reasons that the old lever systems were phased out, electronic systems are fundamentally complex. Also, election judges overwhelmingly are retirees, the least computer-savvy population segment. Should one of these complex machines glitch (printer dies, paper continually jams, a station smokes for no good reason, a display gets too dim to see, etc), they are not as prepared to handle the problem as would be ideal. Enough training will help this, but even optical-scan causes some people problems and that's been around since the 1970's.

No, electronic voting is not a good idea. It sounded neat in 1988 when the futurists thought it up, but reality hasn't caught up with them yet.