Myki, top-up speed, and distributed data systems
Russell Degnan

The release, last week, of both the ombudsman's "Investigation into public transport fare evasion enforcement" and the government response shed some important light on public transport ticketing and the difficulty of both enforcement and compliance. These are long-standing issues, with a culture of intimidation and fear that dates back a decade and more. Moreover, the general problem of fare-collection will only get worse. Policing a system with largely open-access and micro-payments is very labor-intensive and therefore increasingly costly. The "automated" ticketing system introduced with Metcard (actually, the short-lived scratch-tickets) fixed only the selling element of this problem. And in the case of Myki, did that badly as well.

The general problem will continue to exist regardless of the ticket solution. It is predominantly a problem of open-access and there is little chance of that changing on the train tram network, and none at all on trams. But the issue of selling tickets, and the problems Myki faces with updating passengers on their correct balance, or taking online payments can be fixed. It was telling that the government list of measures began with this change:

Reducing the time it takes to top up online from 24 hours to 90 minutes

From a user perspective it is of some but limited value. If a user recognises that their Myki balance is in the negative, then they can top-up in time for the next trip. Commonly though, a user will board a tram not knowing that their ticket has a negative balance: the lack of touch-off for users of only trams means Myki reports the previous balance, not the new one on touch-on. Myki's inability to allow users to immediately top-up their balance runs counter to the experience people have with payment systems in every other part of their life. There is a reason, but the reason relates to the technology Myki was built on. Technology that is now two decades old, built for a different time. One that has now passed.

Web Businesses Will Live in the Cloud - Marc Andreessen

Contactless smartcard systems began to be introduced in 1991, slowly gaining popularity, notably after Seoul and Hong Kong which introduced them in the late 1990s. When Myki came into being, these were the gold standard by which all ticketing systems were expected to meet. Troublingly, despite there being no market imperative to have a ticketing system of this sort, the Transport Ticketing Authority chief executive Bernie Carolan indicated in 2012 that this was also the main reason: "It was known that other transport systems were heading towards smartcard systems, so the thought was, the sooner we do it, the better."

The momentum behind RFID technology for ticketing systems shows no signs of abating, as many more have been introduced in the decade since the Myki project was started. Yet this technology is nearly unique to transport (and other government) authorities. Businesses, handling trillions of online payments, and producing billions of tickets for everything from major concerts and sports events to the self-created and managed sales of companies like EventBrite, and even transport companies like Melbourne's SkyBus use bar-code technology.

The reason related back to Andreessen's prediction, made in 1999 and quickly proven correct. In internet terms, the late-90s are a different world. Storage was expensive - the textbook for data management I used in 1998 was titled "Managing Gigabytes" - and the internet was confined to fixed networks, and dial-up. In a transport scenario for a large city like Melbourne, with half a billion trips per year, storing and accessing a terabyte of yearly travel information, and transmitting user information to roaming vehicles in real-time was unthinkable. Hence the need for distributed data storage in the form of RFID cards. The network is built into the system itself: the card stores the data - the account value and travel information - and is updated by the machines. While the 1990s computer technology Andreessen predicted would move from distributed programs (like Office) to the internet (like Google Docs) has slowly switched, transport technology remains distributed across millions of cards, and the system must maintain thousands of complex readers to reflect that distribution.

Which brings us back to online top-ups. In our modern world, being able to transfer money to an entity, regardless of your geographical position, and seeing an immediate credit is standard (and indeed has been for ten years) The inability of Myki to register a top-up immediately is incongruous. However, while Myki was built in the internet era (though not the wireless one), the underlying technology was not. A monetary transfer to a card needs to be sent to the card reader in order to update the card. Moreover, because there is no way to predict which card reader a user will use, it must be broadcast to all card readers. The data involved is not large - not least because few people use online top-ups when they won't register - but it is clumsy and difficult to manage, and results in other perverse effects.

Most of the "problems" of Myki are built into the distributed system design and the need for (expensive) chipped cards:

  • The most obvious of these is the disconnect between an account and the card. Because the data is on the card, losing it means losing the data, with a complex replacement process of loading old credit onto a new card and re-linking it to the account. If there was no account, the credit is lost forever.
  • Single-use tickets are problematic when cards cost several dollars to create, and aren't just a signifier of trip eligibility as they were under Metcard and earlier ticket systems.
  • Expiring cards - presumably a measure of the lifespan of the chip itself - is a product of the card being more complex and therefore needing to be periodically replaced.
  • The complexity of distributed storage that requires software checks for account changes, partial card updates, and uncompleted trips; and expensive hardware/software roll-outs to the readers when policy changes are made

Distributed systems are rarely used in modern web programming in favour of central storage because data and network connections are now cheap and reliable. In a centralised system the data the user has to have is much simpler: either an account identifier that allows them to be recognised; or a trip identifier, with the relevant date and access encoded.

The problems identified above with Myki would disappear, as a consequence:

  • Online top-ups will reference the central account, which is the same database an inspector will check. A user could receive a low account warning when boarding the tram, and use a mobile app to add credit to their account immediately without even needing to re-touch-on - the current trip counting as pending.
  • Because a "card" is only a bar-code or equivalent identifier, they are easily reprinted (or regenerated if security is a concern), could be stored on a phone display, rather than a paper or plastic version, and extended to multiple users without any interaction with PTV.
  • As already discussed, single-use tickets are paper versions of temporary accounts - or alternatively, real accounts that can be topped up, but printed on paper card, without the expense of an RFID chip
  • Account expiry would not need to occur, though unused accounts might be archived for data reasons, and the lifespan of the card is irrelevant if a user can reprint on demand - or never prints it at all.
  • And finally, centralised software means that all users (and machines) are always working with the most current version. It vastly simplifies the hardware requirements, as the only requirement of a bar-code style system is a display (an embedded browser is sufficient) and a bar-code reader that triggers a network call. There is no possibility of a partial update (as there often will be if a transaction is attempted but the card removed prematurely), as the transaction is operated at the network level. The only concern is unreliable networks - which might be a problem on some trams or buses - but it is a problem that is rapidly disappearing, and one that can be solved with cheap hardware investments.

With the contract for Myki coming up for re-tendering, the government is caught between two unpalatable decisions: continue using a system that is widely disliked, works badly (even by the standard of contactless smartcards) and by design includes incongruous limitations that are not easily fixed (if at all); or they can invest in a system that reflects modern programming and technology, but equally worryingly, be at the bleeding edge of transport ticketing systems (if still behind the curve in industry terms). The upsides to the latter approach include the possibility of a vastly superior system that is much cheaper, and the opportunity to on-sell it to other cities if successfully implemented. The downside is the possibility of another multi-billion dollar IT ticketing disaster to follow one from a decade ago. A business with Myki would be haemorrhaging customers and need to invest in better systems. A monopoly provider of ticket services can afford to be terrible. Its failures can be passed on to customers as fines for non-compliance, or the government in lost revenue; its added maintenance costs are borne by tax-payers. There is no easy choice, but it would be rare to find a user of Myki who wouldn't prefer something vastly superior.

Sterner Matters 4th June, 2016 20:52:20   [#] [1 comment] 

Optimal stop spacing and travel distance
Russell Degnan

In my previous post on stop spacing I made the point that for short trips - particularly the last mile shortening stop spacing counter-acts the benefits of shorter walks because the transport slows down. Efficiency and speed is currently under-rated - in comparison to connectivity and frequency - in transport discussions (at least in Melbourne). It shouldn't be, for two reasons: speed is the primary determiner of the route and travel method chosen; and faster transit leads to faster turn-arounds and therefore fewer trains and drivers for the same frequency. A 20% improvement in travel speed doesn't increase capacity (that remains throughput) but it would have massive implications for recurring costs.

In this post I'll discuss trips of varying lengths, in order to make a simple but important point: in a walk-transit-walk environment optimal stop spacing is a function of travel distance. A secondary point will also be made, with caveats: that for many systems, particularly in Melbourne, stop spacing is much too close. (A point made in relation to trams in Melbourne by Jarrett Walker)

Firstly a few assumptions. Adjusting them may make some small differences - and if anyone wants the spreadsheet I did this on, just ask - but less than you might think. For the sake or argument I am assuming a grid with even density, so average walking distance is equal to stop spacing at both ends of the journey: half the users will walk less than half stop spacing, half will walk more than half, with those in the centre of the grid traversing half in both a N-S and E-W direction. The transport in question has a 1m/s2 acceleration and deceleration time, with intersections ignored (ie. light-rail, either grade separated or gated), a 40 second stop penalty, 5 minute waiting average, and a walking speed of 5km/h. Graphs will reflect averages; walking, transport and waiting time will vary, obviously.

But to emphasise, again, adjusting these numbers makes very little difference to optimal stop spacing: waiting time is a constant, and only matters if someone can walk the distance faster - ie. for very short trips. Otherwise walking and transport speed are minimised at the point where transit speed is not compromised by frequent stops: the major factor in determining optimal stop spacing is the distance being travelled on transit (given a particular walking speed)[1].

Optimal stop spacing in a walking only environment

Below is the journey speed for a 4km trip, given the assumptions above. There are two things worth noting. Firstly, that optimal stop spacing is 850m, which is outside the generally accepted range for short-ish trips of this kind. Secondly, that the cost of sub-optimal spacing is much higher on the short side. A 400m stop spacing is of a piece with a 1700m stop spacing, and the cost of reducing it further much higher.

There is an important, and unresolved tension then, between two conceptions of walking to transit: is the commuter rational, and therefore willing to walk whatever distance affords them the fastest trip, where speed is all that matters; or are they unwilling or unable (speaking here of the general population, not the mobility impaired - and of a decent walking environment, as that can be fixed) to walk further, even if it meant faster transit? This graph from VISTA data (courtesy Alan Davies), would indicate that people are willing to walk reasonable distances for trains (which on average have longer travel distances), but may merely indicate that many train stations are further apart. Similarly, while there is a significant clustering effect around train lines in Melbourne, that is in part because the stop spacing is short, and therefore geared for short walks.

The graph below shows the optimal stop spacing for various trip lengths. The levelling out at 2km is a function of the maximum travel speed of the transit, but shows that except for very short trips the optimal stop spacing is in excess of 1km, and growing. Keeping in mind the previous point that longer is better than shorter, a large percentage of commuters in Melbourne are suffering excessively long travel times.

Optimal stop spacing with a connecting transport

While Melbourne has an high percentage of walking only access to trains, this is probably a reflection of both poor local connections, and the short stop spacing that allows extensive walk-up access at the expense of travel speed. I'll cover the relative speed of Melbourne public transport in a future post. In large cities, with long commutes, feeder systems allow for a slightly different stop spacing arrangement, because they can cover for shorter trips, and allow the train system to focus on longer trips at higher speed.

In the following graphs, a feeder system, travelling at an average speed of 30k/h[2] with a 400m stop spacing takes passengers between stops with a 5min wait for a connecting service. The difference this makes to a 20km trip can be seen below:

Again two points are worth taking from this graph. The first is that a significantly longer stop spacing is optimal when there is a connecting service. The second is that the margin of error for the stop spacing on the long distance service is much higher. Anything from a 3km to 8km stop spacing gives a broadly similar transit speed because the service is mostly running at full speed. This is important, because transit must, of necessity, serve trips of different lengths. If the speed is broadly similar regardless of the stop spacing then (provided the basic minimum stop spacing is achieved) transit agencies can place stations in major centres to maximise connectivity.

The graph for optimal stopping distance across all commute lengths shows how much further out the stops can be with a good connecting service.

This conclusion is in some ways obvious: naturally transit can go much faster if it doesn't stop, and naturally systems that interact will work better. But it leads to several important points:

  • It is better to have stop spacing too long than too short, because the time penalty is significant.
  • Transit for long trips and short trips is not interchangeable. This is particularly true if transit designed for long trips has better frequencies, as it will out-compete local transit, making it redundant.
  • A system that allows bleed between the roles of different transit will be sub-optimal. Part of long term planning should be to optimise the system for efficient travel by reorganising stops and connections.

Needless to say this has important implications for Melbourne's public transport system.

[1] Here I have assumed walking is a constant, but note that even at very slow speeds (1km/h) the optimal stop spacing for a 10km trip is 650m, compared to 1300m for a walking speed of 5km/h.

[2] This would be an unusually fast service given that level of stop spacing, and most likely, a street running route. Halving this to 15km/h reduces optimal stop spacing for the rapid transit service by around 15%

Sterner Matters 26th May, 2014 01:20:55   [#] [2 comments] 

Mode Choice and Rational Commuting
Russell Degnan

Amongst the various comments about the proposed Metro line and the new station at Fishermans Bend, certain forms of analysis stood out, not for their accuracy, but their falling back on cliched understandings of travel, and the limitations of simple analysis in complex areas.

There are issues with the new proposal, mostly around the absence of long-term strategic planning in favour of project lists, which is not a new complaint.

There are fewer issues with a station at Montague, notwithstanding the commentary suggesting otherwise. Jason Murphy is normally very good, but his use of coverage diagrams merely highlights their limitations. Coverage is only as valuable as the service being connected to, the services that service connects to, their frequency and the line speed. Moreover, residential coverage is meaningless in the context of an area with several major destinations (the conference and exhibition centre, and South Wharf) and the expectations of significant future employment. What matters is the way that area connects as a destination. A connection to South Yarra and Southern Cross is potentially very useful; if debatably similar to the proposed (if unlikely) connection to Newport and Flinders Street.

The complaint of both Daniel Bowen and Tony Morton that residents would not use the station to connect from the light rail is in a similar vein: correct but basically irrelevant, if the stop is considered as a destination.

But they highlight a more general transport problem that is worth noting and explaining:

Public transport has big trade-offs for short trips

This often comes under the problem of the last mile, whereby a trip that ends at a transport hub needs a short connector that is hard to serve efficiently. But the problem of serving the last mile is true generally for short trips.

Consider someone within the residential area of the proposed Montague station, working in the CBD. In general, the potential user won't be either next to the residential station or working at a station in the loop. For the sake of argument we'll put them 400m from each.

As a walk, it is 3km into the centre of the CBD, or around 30 minutes. A trip to Docklands would be shorter, other parts of the CBD potentially longer.

As a train trip is is two walks of 400m (or 8 minutes), two trips through the station and onto underground platforms (2-4 minutes), waiting for the train (2-5 minutes), and a 2km trip on the train to Southern Cross (3 minutes). Somewhere between 15 and 20 minutes, most of which is spent walking and waiting.

On any transport, the walking distance can be shortened by reducing stop spacing. But here again, there is a trade-off in travel time as the stops add minutes, and the transport spends more time accelerating and decelerating at lower speeds. A tram with a 200m stopping distance would quarter the walk on average (2 minutes) but stop six times (5 minutes), and take 6 minutes to make the trip at the lower speed. Making the trip, including wait time, between 13 and 18 minutes.

In his piece on Fishermans Bend Alan Davies sourced a graphic showing walking lengths very substantially by mode in Melbourne. Most likely, this reflects two things: firstly, that stopping patterns are shorter for buses, and therefore the walk is naturally reduced; and secondly, that, although the environment has some effect on preferred walking distance, commuters are largely rational with respect to walking distance and time. Train journeys are longer, and walking is more efficient than the poor connections available in suburban Melbourne (on which, more later).

That being the case, the "rule of thumb" noted by Jarrett Walker of stop spacing between 400 and 800 metres is flawed. If we assume for rational commuters trying to minimise time, for a given trip length, there is an optimal stop spacing where walk time is offset by the speed of the transport. Because commuters go vastly different lengths this distance actually varies, and is often substantially longer than 800m. Though as a future post will show, this is complex; for large metropolitan areas, connections matter, a lot.

But on a short segment such as to Fishermans Bend, where most CBD/Docklands or local trips will be between one and four kilometres, there is no time advantage to having a heavy rail line with stops a kilometre apart through the area. The maximum possible trip is a mere 7km, which just barely goes past 1km as an optimal spacing. Stop spacing of a kilometre is double the optimal length for most of those trips. Yet on heavy-rail, any further shortening of those stops is impossible, and would both: effectively make the service a very expensive light-rail line; and were there a Newport connection, significantly slow suburban passengers.

The graphs below show the various trade-offs, although they slightly over-simplify the longer stopping lengths as at some point it becomes quicker to just walk, and wait-time is eliminated. [1] Notice too, that the optimal stopping distance is, as expected, between 400 and 800m for trips in this range.

There may be a future capacity issue in Fishermans Bend, given the projected population, and there would be a case for a station in Wirraway that connects to local transport in the event the South Morang-Newport connection occurred. But the future residents of the suburb will be much better served by efficient (and substantially cheaper) light-rail/tram lines. The vast majority of their trips (and the only ones reliably performed by public transport now) are too short to gain anything from services better designed for much longer journeys.

[1] Some assumptions need to be noted: the transport in question has a 1m/s2 acceleration and deceleration time, with intersections ignored (ie. light-rail, either grade separated or gated), a 40 second stop penalty, 5 minute waiting average, and a walking speed of 5km/h. It reflects averages; the transport time and waiting time will vary, obviously.

Sterner Matters 10th May, 2014 19:40:19   [#] [2 comments] 

The crazies are here; early Senate wrap 2013
Russell Degnan

Yesterday I posited that this could see a historically significant shift in minor party support that raised the prospect of any number of unknown parties getting into the Senate. This morning I updated the scale on my chart because Nick Xenophon destroyed it, and we are all researching the policy platform of the LDP, Motoring Enthusiasts, Sport Party. Palmer United and Family First comes across as mainstream by comparison. Here is the updated graph:

What is clear is that Labor has been wiped out in the Senate. This is catastrophically bad for them, and it will take a fundamental shift away from the minor parties for them to approach the numbers they'll need for third seats in future elections. The Liberals have no need to cheer either, as they lost ground in all but two states and saw no significant shift towards them even there. This election might be a once-off, but the trend for the past 30 years suggests it marks the beginning of a polyglot in the upper house.

One other thing to note: it is absolutely true that some parties are going to get in on the back of dubiously small first preference votes. But don't be deceived into thinking that - with the possible exception of NSW - they would have preferred Labor or Liberal or Green. If they'd preferred that they would probably have voted for them; and they clearly haven't. People may not approve of the particular minor parties, but these are not the first minor parties in the Senate, and the non-major-party vote continues to grow. People absolutely want someone other than Labor or Liberal (or the Greens) represented in the upper house, and they have certainly got it this time.

Which is not to say that the Senate voting system of 1 above the line doesn't need changing. It clearly does, and even first past-the-post with multiple members would work better than the cluster-fuck the major parties created thinking they could use it to shaft the Democrats. Getting reform through a Senate created by the process though? Best do it before next July.


South Australia
Nick Xenophon might regret not putting in a harder line of preferencing for his running mate - or more effort into polling. Two full quotas have been added to the minor party vote despite a decline from the Greens. Labor are nowhere near a second quota and will fall behind the Greens, elevating Hanson-Young back into the Senate. For reasons best known only to them they then elect Family First ahead of the second Xenophon candidate, and while they'd have got in anyway, based on Labor's preferences (amongst many others), those 10,000 votes will be decisive in electing the Liberals over Stirling Griff, unless there is a huge shift in below the line votes.
Liberals 2, Xenophon 1, Labor 1, Greens 1, Family First 1

Western Australia
The first sign of the Labor Senate apocolypse. The Liberals and Nationals combine for just over 3 quotas. But all the fun is in the Australian Sports Parties Steven Bradbury-esque miracle run through the exclusions. It is hard to see this holding through the below-the-line votes and pre-polls/postals. They start 1200 ahead of last, escape (and harvest the preferences of) the Climate Sceptics by 260 votes, Rise Up Australia party by 200 votes, the Democrats by 640, and depend on a couple of other key exclusions to ensure neither the Liberal Democrats nor the Palmer United party get up instead. If it falls as written, Palmer United will elect the Greens over Labor for the last seat. Weirdly, WA fell in a more sensible spot: Liberals exactly three quotas, Labor and Liberals deep into the count. But there is more than a quota of right-leaning micro/minor parties and that means one will probably get up.
Liberals 3, Labor 1, Greens 1 Sports Party 0

Queensland
A relatively sensible count unless Labor drifts further in which case a few other scenarios are possible. Both the Liberals and Palmer United will get up on Family First, Katter or Fishing votes, and the Greens vote, while big enough for second, can't get close to another quota with Labor in such dire straights (and them for that matter, remembering they took 6 seats three years ago). Not that Labor was helping them in Queensland anyway.
Liberals 3, Labor 2, Palmer Utd 1

Victoria
A Senate ballot and count so confusing I am not entirely sure who I ended up voting for. Labor and Liberal both cleared two quotas, and the Labor vote will shoot the Greens into a seat with a bit to spare. That least seat though, got a little weird. The Motoring Enthusiasts, with only 0.57% of the vote seemingly survive to the end, starting with the Fishing and Lifestyle exclusion (2600 votes), and harvesting regularly with only Family First (17000 votes) and Palmer United (15500 votes) putting them in doubt. We won't know for sure until they push the button, but with such large leads this is almost certainly an interesting result.
Liberals 3, Labor 2, Greens 1, Motoring Enthusiasts 1

New South Wales
Speaking of interesting. What ought to have been a straight-forward vote got weird because people in NSW made an even bigger blunder than Victoria did when the DLP got up. There is no other way to describe the Liberal Democrats getting 8.9% of the vote from the first slot while the similarly-named Liberals shed votes. There are no close exclusions in the count, and they are near certainties to be elected. For what it is worth, swing that marker around and the Liberals take three quotas promoting someone else - possibly Palmer United Party, possibly Shooters and Fishers. It seems unlikely that the Greens would have got in regardless. The combined Labor/Greens vote is too small, and they need to make up 1.5% to take the 6th spot; if we are honest, the LDP might actually add something to the parliament, unlike certain others.
Liberals 3, Labor 2, LDP 1

Tasmania
It turns out the swing against Labor was even bigger than expected, and a strong swing against the Greens meant they had to depend on Labor preferences for that final spot. The Liberal Democrats go deep in this count with 2% of the vote (and are within 600 votes of Palmer United at the final exclusion), but none of the last six exclusions preference them, so they'll get second at best. Palmer United walking home on the two-party preferred; although Tasmanians are known for below-the-line voting, so if anywhere can throw a spanner in the works it is here.
Liberals 2, Labor 2, Greens 1, Palmer United 1

Australia Capital Territory
This might not be known until they push the button. The Liberals are under on first preferences, but only by a couple of thousand votes. But with the Greens getting all but a handful of those preferences, they are more likely to fade, than catch up with the extra large below-the-line vote in ACT. But given the lead is projected at 1200 votes, the composition of postals and pre-polls could swing things around yet.
Liberals 1, Labor 1

Northern Territory
Labor missed a full quota, but get past the line with Sex Party votes - if that matters. By that stage Palmer United had forged ahead of the Greens anyway, meaning Nova Peris would get in on Greens preferences, even if the count changed.
Liberals 1, Labor 1

Assuming nothing changes - though WA remains very much up in the air - the Liberals will need to find six votes from either the Greens (10), Labor (25) or a coalition that almost has to include Clive Palmer, plus some combination of the DLP/Family First, Nick Xenophon, the LDP, and what amounts now to a sports lobby. If Palmer can take the final WA seat he'll effectively control the Senate. Even Barnaby Joyce is scared of that. Victorian Secession never looked so good; at least we only elected a motoring club and the Catholics.

Sterner Matters 8th September, 2013 18:31:51   [#] [0 comments] 

The crazies are coming; Senate predictions 2013
Russell Degnan

Polls close soon, so best get on to a now recurring tradition of incorrectly predicting the Senate outcome. Actually, in terms of reading the polls, 2010 was pretty close. Looking back and comparing to the image of what actually happened shows that, unlike in 2007, the problem was under-estimating the swings in WA and Queensland, underestimating the increase in minor party support in Victoria and NSW and over-estimating the swing in NSW and Tasmania. South Australia was practically perfect.

Translating that into seats worked, for the most part, but not always for the right reasons, and the DLP victory in Victoria came somewhat out of left field given the incumbent was Family First - though both sailed in on preference deals, not merit.

Unfortunately, that scenario, where a party rides a preference wave is becoming increasingly likely. That makes calling the Senate a mugs game, because although you an make reasonable predictions about the major parties, the rest depend on the fluttering of butterfly wings. Most importantly, with the minor party vote now consuming more than two quotas in many states, Labor and Liberal become king-makers for whoever is left over.


Anatomy of a Preference Deal

Antony Green has already said everything that needs to be said about the proliferation of metre long Senate ballots and the perverse nature of deals creating Senators from tiny votes. But it is worth dwelling for a second on why this works.

If you are a minor party your aim is to stay in the count; that means harvesting preferences from parties smaller than you, to get bigger than parties bigger than you, then doing the same to them. There is never any value in trading with the Greens (inevitably nearly a full quote and one of the last removed from the count) and only late in the count with Labor or Liberal. The aim early on, promise, and effectively combine your vote with someone else. Thus, if two parties have 0.5% of the vote, and deal with each other, they effectively become one party with 1.0% of the vote. Do that ten times and then target the mid-size parties - the Palmers, Katters, Family Firsts, DLPs and Liberal Democrats (if they sit left of Labor/Liberal on the ballot). With the major party vote dipping below half a spare quota, a combined vote of 7% means any late preference deal with them can propel a party to a seat.

In a few cases in this election, the right party, in the right place, at the right time, can get up with less than one percent of the vote. With luck, leakage from below the line voting will prevent this, but how much leakage can occur when voters are filling in 100+ boxes? It isn't even clear to me where my vote might eventually end up, because it depends on subtle sliding-doors points in the count.

For this reason, take what follows with a grain of salt. Minor parties can definitely claim seats (for good or ill, some variation doesn't really hurt), but predicting it is nearly impossible, even with the sort of monte-carlo simulations Truth Seeker is using. What I'll highlight here is where certain scenarios become more likely, and why, if you are following the count late into the evening.


2013 - Historically Significant

Track back over the minor party shifts in the last 30 years and one thing becomes clear. They are capturing a lot more of the vote. In the mid-80s Labor and Liberal were taking 6 quotas by themselves Now they are struggling to put together five. There is a magic line at 72.85% combined first preferences for Labor and Liberal that takes us into the unknown. Above that, and the Greens will generally get one spot, and the minor parties elect the stronger of the two others (normally the Liberals). Below it and there is enough of a quote remaining for the minor parties to pull a quota without the Liberals help, leaving the stranded in the count.

The shift to this territory is new. There is no guarantee the shift will happen either, as the strength of minor party votes are unpredictable, even after polling. But the historical trend, and the polls says it probably is happening. In the last few elections the first-preference major party vote in the Senate has been 4-7% below the house of representatives. Combine that with the latest polling: welcome to Crazy Town.

Let's see what we have. Primary votes aren't mentioned often enough. Obviously two-party preferred matters for the house but the Senate tracks nearly perfectly along first preferences, with an adjustment for minor party losses. And what losses! Newspoll predicts a 4% increase in minor party votes over 2010; the BludgerTrack is, if anything, even stronger. Given they were already historically low numbers this takes at least three states into unprecedented sub-70% mark.

South Australia
Nick Xenophon is back to reverse the strong shift back to the major parties in the last election. Even if he fails to meet a quota he is helped by the Palmer United party and the Nationals. The Greens, by contrast, might be struggling if Labor fails to get far enough over a second quota to push them over the line. This ought to be straight-forward Liberal 2, Labor 2, Xenophon and Green. The No Carbon Tax Climate Sceptics have a ridiculously good preference run, winning often from almost impossibly low positions. Watch that combined Labor-Green vote.
Liberals 2, Labor 2, Xenophon 1, Greens 1(0) NCT 0(1)

Western Australia
Western Australia is pretty much seceding at this point. Their vote went strongly to the minor parties in the last election, but that looks less robust this time, with a slight shift to the Liberals. As in SA the Greens need to get very close to a quota AND hope Labor passes a second quota, neither of which are certain. If not Palmer United (especially given recent polling), the Liberal Democrats (with a good ballot position) or who knows are all in with a shout.
Liberals 3, Labor 2, Greens 1(0) Palmer Utd 0(1)

Queensland
As in WA, I am not convinced the minor party will increase by a lot, but it was already running 10% better than in the HoR, so increasing to around 30% is possible. From that position, anything can happen. Labor is not preferencing the Greens first, AND has very little over a quota which could lift Katter or the Stop CSG into a senate seat. That's if they get a second quota, which they might not, given how close to the line they are. Palmer United look well placed to stay in the count and win a seat, but their vote is not that strong, and this could go any number of ways.
Liberals 2(3), Labor 2, Greens 1(0) Palmer Utd 1(0) Someone Random 0(1)

Victoria
If you thought the above was bad wait until neither Labor nor Liberal have close to a half quota and drop out before the important bits. Labor's late vote collapse - if polls are accurate - has probably rescued the Liberal's third candidate from being overtaken by the right-wing party, or someone random. But this presumes a fairly substantial swing to Labor and no substantial swing to the minor parties. A drop straight down will throw up any number of odd scenarios; though fortunately a number of parties lack the competency to hand in a group-ticket form, which cut down the number of random results markedly.
Liberals 3(2), Labor 2, Greens 1, Family First 0(1)

New South Wales
This should be fairly straight forward, as the Liberals are close to three quotas, and Labor will push the Greens over the line. The major party has held up fairly well in NSW in recent elections though, making it somewhat higher than elsewhere. If the voters come out with baseball bats for Labor then their remainder might not elect the Greens and with the mega-ballot-paper anything can happen (see Queensland).
Liberals 3, Labor 2, Greens 1

Tasmania
Let me quote from three years ago "who knows what is going on here". Lack of polls doesn't help. Are we really looking at a 12 per cent swing? And if so, is it to the Liberals or to others? Tasmania has a strong tradition of voting below the line, which is good, because a couple of parties (the Independents) have phenomenal harvesting runs. The Greens should pass their first quota easily, but as above, that last seat could go anywhere as the major party vote dips below 70%.
Liberals 2(3), Labor 2, Greens 1, Family First 1(0) Someone Random 0(1)

Australia Capital Territory
Always the easiest count, but with a really high below the line vote, and personal connections to candidates there is always the chance that the Greens will finally tip the Liberals below the 33% quota and them in. It seems unlikely in an election favouring the Liberals, but some think it is possible.
Liberals 1(0), Labor 1, Greens 0(1)


Northern Territory
Normally a ridiculously straight forward vote, the complete collapse of Labor support has both significantly increased the minor party vote and made them vulnerable to minor party challenges. I've no way of verifying if Labor will drop below 33%, but this is at least interesting for once, as explained here.
Liberals 1, Labor 1(0), Greens 0, First Nation 0(1)

By the numbers:
Liberals 17(17), Labor 14(13), Greens 6(5) Others 3(5)

Continuing:
Liberals 16, Labor 13, Greens 6, Others 1

In total:
Liberals 33, Labor 27, Greens 12, Others 4

If the Liberals have a very good election they might just avoid having to negotiate with the Greens or Labor, in favour of a collective mostly right-leaning group. But that seems unlikely, and the balance is sure to lie with the opposition. Which, incidentally, with over 30% of the vote going to minor parties, even if not to just that minor party, might be how it ought to be.

Sterner Matters 7th September, 2013 18:43:22   [#] [0 comments] 

Terrible driving, poor riding, but not distracted
Russell Degnan

I really shouldn't be as annoyed by the advertisement as I am. The design is clever and attractive; and the message is straight-forward and not really in dispute; walking or cycling without your hearing sense (or visual sense if you are looking at your phone) makes your day unnecessarily dangerous; although in-car distractions and the sealing off of drivers from the outside world continues apace.

But the depiction of the cyclists sends terrible messages about cycling and driver responsibility, for three reasons.

Firstly, a cyclist should never be riding in amongst the parked cars, as it requires them to continually merge into traffic and makes them harder to see. Veering out into traffic like that just shouldn't happen, and the depiction of a cyclist in that situation sends a message that that is where the should be.

Likewise, a car coming up from behind a cyclist, travelling down the road, should have the awareness that they will move out and around impending obstacles. By putting the blame on an accident of this sort on "distractions", it simultaneously implies that the undistracted cyclist needs to stop before going around a parked car to wait for oncoming traffic.

Whereas an accident of this sort has nothing to do with distractions, unless it is by the driver. Rather the cyclist ought to ride in a manner that maintains their presence and visibility in the driver's zone of attention; and the driver, regardless, needs to be aware of the presence of the cyclist and give them space. To imply that the fault in an accident of this sort would fall on the cyclist is plainly wrong. Cyclists can be distracted, but riding through a give-way sign would have been an accurate depiction of an accident where a distraction and not poor driving was to blame.

Sterner Matters 6th August, 2013 01:25:35   [#] [0 comments] 

Senate Predictions 2010
Russell Degnan

Somewhat belatedly, I thought it would be worth revisiting the Senate prediction method used in 2007. Unfortunately, it wasn't as effective as hoped, as can be seen by the graph below.

While 1996 saw a strong movement away from the sitting government, and towards minor parties, the 2007 election was closer to 1983. Strongly towards Labor directly from the Liberals, rather than the collapse of the government vote.

Thus, while the swings were pretty close (although consistently overshooting), the minor party swing was the reverse of that expected, except in South Australia (where the Xenophon effect had been extensively polled) and Tasmania (which is seeing consistent growth in the Green vote).

This election promises to be more typical, following the patterns set in 1984 and 1998 where first term governments suffered heavy swings without any increase in the opposition's primary senate vote.

Tracing the HoR polls, as before, onto primary vote percentages, and the expected minor swings gives the following predictive graph.

Using Antony Green's senate calculator produces the following in each state:

South Australia
Unusual, in that the lack of a Xenophon votes means the minor parties will probably go backwards, to roughly in line with historic trends. Two seats are up for grabs, leaving the three main parties in a race. The Greens need over 11-12% of the vote which is puts them on the knife-edge, given recent polls. The Liberal easily ride to 5th, but Democrat preferences will go to Labor, so the Greens lot depends on whether their vote has a post-collapse anti-major electoral rebound, but also, by how much the Liberals pass their third quota, as the Greens get almost nothing from anyone until they are out.
Liberals 3, Labor 2(3), Greens 1(0)

Western Australia
A straight-forward count, giving the 6th sport to the Greens easily, the CDP makes a late run as it scoops up votes from all-comers, but the Labor party will tip the Greens over the line, without making a serious challenge for the sixth spot themselves.
Liberals 3, Labor 2, Greens 1

Queensland
A straight forward vote, Liberal pick up an easy 3 from the right wing vote. Given the expected swing, Labor will never challenge for the 6th, put use their preferences to put the Greens over the line against Family First, who should slide into second.
Liberals 3, Labor 2, Greens 1

Victoria
Greens easily take 5th, as they have Democrats preferences if they fail to make quote themselves. Any pro-Gillard swing will put Labor very close to a third quota, but they get almost no preferences above the Liberals, so the Liberals need around a 37% primary vote to crawl over the line. If the Liberal vote goes further south, Steven Fielding will need both around 4% himself and a weaker than expected Green/ALP vote to take advantage of the DLP's pre-ALP ballot position and crawl back into Parliament. Unlikely.
Liberals 2, Labor 3, Greens 1

New South WalesInteresting count. Labor's likely vote collapse means they need both the Green/Democrats vote to stay below a quota and their own vote to stay above 36% to come second above the CDP. If they succeed, right-wing votes will tip them into a third quota, otherwise the Greens will carry Labor preferences home: they need around 11.5% to win, right where they are currently polling.
Liberals 3, Labor 2(3), Greens 1(0)

Tasmania
Who knows what is going on here. Most likely outcome is actually 6 quotas filled and no distribution. But there is an outside chance that both Labor and Liberal will fail to fill their 2nd/3rd quotas, leaving the Greens or Liberals a chance of taking the 6th seat if their vote explodes upwards.
Liberals 2, Labor 3, Greens 1

Australia Capital Territory
The Greens always hope to win this one, but they need a strong Labor vote to keep the Liberals under the quota of 33%. That's unlikely amongst a big NSW anti-Labor swing, and almost impossible with Democrats preferences heading towards the Liberals, though a pro-Labor, anti-Liberal population turnover means it can't be ruled out completely.
Liberals 1, Labor 1, Greens 0


Northern Territory
Forget it.
Liberals 1, Labor 1, Greens 0

By the numbers:
Liberals 18, Labor 16(18), Greens 6(4)

Continuing:
Liberals 16, Labor 16, Greens 3, Xenophon 1

In total:
Liberals 34, Labor 32(34), Greens 9(7), Xenophon 1

Sterner Matters 21st August, 2010 19:30:03   [#] [0 comments] 

Women and cycling
Russell Degnan

Much is made of the large disparity between the number of male and female cyclists in some countries - notably the USA, UK and Australia. Numerous differences are cited, from appearances to different patterns of use - most of which don't stand up to scrutiny given the higher rates of female cycling in many northern European countries.

Melissa Lafsky at The Infrastructurist cites some physiological reasons, arguing that commuting by bike is more amenable to the male, testosterone driven mind-set. This is, by no means, a unique observation, but it may be an important one. If female risk aversion is the major reason for lower cycling rates on unsafe streets then we should see some differences in commuting patterns.

Unfortunately, strong data on commuting patterns and demographics is hard to come up with, particularly at the sort of finely grained detail needed for this type of study. A student of mine looked into this earlier in the year and observed almost double the percentage of female cyclists on roads with bike lanes (45%) versus roads without (25%). Unfortunately the study was also small, short and biased towards routes that we already know from the census had high numbers of female cyclists - which may or may not be a cycling lane thing.

Still, this was valuable confirmation of the idea expressed above. The more recent availability of CData for the 2006 census allows me to test another hypothesis: if female cyclists are more risk averse, then the percentage of cyclists that are female should correlate strongly with the percentage of cycling commuters.

Using data from every Statistical Local Area in Australia, we can see graph these two data sets to see what occurs. CData's tendency to randomise small numbers makes this graph a little problematic. Quite a few SLAs have fewer than 10 female cyclists, and I excised SLAs with either no female cyclists or less than 20 cyclists over-all.


The correlation is not linear, so the graph has a logarithmic scale on the y-axis. The correlation is strong, however. Essentially, for every doubling of the percentage of cyclists commuting, you get a 10% increase in female participation rate. The vast bulk of SLAs have very low female participation rates (5-20%). But there are a number of area (notably in Melbourne's inner north) with both high number of cyclists and close to parity in terms of female participation rates.

I agree, therefore with the conclusion made by Melissa:

"All of which leads to our point (we're getting there, we promise): Letís stop talking about the "women on bikes" issue as a psycho-socio-gender phenomenon, and start talking about it as a policy call to action. If we reprioritized public and private initiatives to push biking, by creating more safety features like mandatory bike lanes, bike checkpoints and safety checks, as well as more incentivizing programs from employers ("bike to work" payment vouchers, etc.), we might see a real and meaningful change in the number of women - and men, for that matter - who chose to bike."

With one caveat. Because women are less likely to cycle when conditions are not favourable they are a better barometer than men if you want to find out why 90+% of the population do not cycle. While there are no shortage of commuting cyclists who have grievances - albeit often important ones - with the policy focus on bicycle facilities, their confidence in traffic and tendency not to expect the same of others is less useful if your aim is to promote and expand the base of cyclists. The non-cycling commuter, particularly the female non-cycling commuter needs to be heard.

Which brings me to my final point. While it was good to see a cycling strategy released earlier this year that actively promoted the idea of cycling as a "serious transport mode", the actual actions proposed, beyond the basic infrastructure already mooted, were thin on the ground. One of the things Copenhagen does very well - largely ignored by politicians who'd rather take pictures of bike lanes on overseas junkets than read a strategy document - is set a series of benchmarks for cycling safety and perceptions of cycling safety in the broader community (that is, outside the existing cycling community as well). We need, in Victoria, proper annual surveys, not of cyclists, but of non-cyclists, particularly women, with regard to their reasons for not cycling, with the aim, through the existing programs, of attacking those reasons. Without that, we are, unfortunately, still aiming in the dark, sometimes at real targets, and sometimes, not.

Sterner Matters 31st December, 2009 21:54:53   [#] [2 comments] 

Assume utopia, then plan
Russell Degnan

Must be a slow news period, as the last fortnight has seen a jump in planning articles in the papers, without there being terribly much news, as such. The first, by Sally Capp, rolls out the usual tropes: suburban sprawl is bad and must be contained, with the solution being greater densities around activity centres - now reduced to six.

Some of the claims are strange, such as the need for investment to encourage businesses to move outwards to the suburbs, despite that being a trend for well over three decades with the actual percentage of CBD bound long distance commutes in new suburbs being as little as 10 percent. But the real problems lie in the author's certainty. Apparently,

If we could agree on the future direction of our city - the proposed activities centres and greater urban density along existing transport routes - we could all be more constructively involved in these discussions.

Which is probably true, but it never seems to occur to proponents of this model that people do, and will continue to, disagree with it. That their inability to make a case for it, not to themselves, but to the people it will affect, is in fact part of the problem. Instead the finger is pointed at local government, apparently a road block to development and lacking in resources to plan effectively. Although how body could plan effectively, lumbered with a planning system that requires extensive consultation, in-built uncertainty and an appeals process that potentially ignores all that came before is not explained.

The idea that a metropolitan body, as proposed, could work any better than local councils and the planning department (who must surely already have all the power vested in a metropolitan planning body), unless the planning system itself is substantially reformed is laughable. That people take it seriously as a solution, without any adequate explanation of how it will improve the system, is sad.

Actually arguing for a strategic direction, rather than merely proposing one never seems to occur in strategic planning discussions however. Which is why clearly inconsistent statements, and proposals can be discussed, with barely an acknowledgment of the other. Take infrastructure. A central plank of the activity centre proposal is that:

By building around existing economic and social infrastructure, we leverage existing facilities without the need to create new ones.

This has been questioned on a number of levels over the years, but I've yet to see an actual economic study showing how true it is, and under what circumstances, for Melbourne, given different strategic plans. Which is why, last week, Frank Keane could write:

About half of Australia's population is contained in five state capitals. The result is an over-urbanisation that is inefficient and requires the building of ever-expanding infrastructure, including transport, sewerage, water and energy supply, telecommunications and waste disposal.

Smaller cities are then proposed as a solution, clearly at odds with high density growth within Melbourne. The relationship of either solution to the economic processes that underpin urban form is never mentioned, so some sense of what is better, or even what might be possible is unknown. The utopian vision for an environmentally sustainable city, or cities, never seems to ask what the point of a city is, before trying to change it.

Developers are on firmer ground, they, at least, understand that a city is there for them to profit from, even in this uncertain climate, but it never hurts if you can get a helping hand. It suits developers to blame local councils for the slow planning process, particularly when there are jobs at stake. Which is not to say the minister is wrong to call in these proposals - though he may be, who could tell? Merely that the existence of these call-ins points to greater problems with the planning system.

Fear and uncertainty prevails in Victorian planning. People don't trust developers, and they don't trust planners. The market has been tending towards the things planners want - polycentric cities, denser development - for years. But the planners reflex assumption that they must constrain the market, and
"encourage" density lends them, with only the flimsiest (and vaguest) of arguments in favour of the plans being created, has turned the planning process into a tool for conflict, with little upside in terms of better outcomes. When times were good, and a little hindrance of development was able to glean the edge off the most abject developments, such a system was poor, but acceptable. When the state government needs to see development, the flaws are more apparent, and planners need to start thinking about what they can justify, what actually matters, and what can be expeditiously jettisoned.

The last activity centre policy failed before it crashed on the rocks of the 90s recession and the Liberal government. The probability of this one following the same pattern are very high.

Sterner Matters 24th April, 2009 18:44:51   [#] [0 comments] 

Data semantics at thirty paces
Russell Degnan

Disputes over the value of Melbourne 2030 are always interesting, some people claim it is effective and producing bad outcomes, some that it is ineffective and not stopping bad outcomes, and some, notably many of the authors printed in People and Place, strike a middle ground that claims it is ineffective at producing things they like, but effective at doing things they don't. By contrast, the government has generally claimed that Melbourne 2030 is doing more or less exactly what they expected, which just happens to be not very much.

The differences in opinion, lie in the interpretation of the actual nature of Melbourne 2030, and its claims, and the expected changes to the urban form in the absence of any plan.

Because dwellings are built almost exclusively by the private sector, Melbourne 2030 is not a facilitator of anything. Even when the legal planning framework is twisted around to claim that no developer should expect to be able to develop without local approval, the developer is still the deciding factor in whether a development goes ahead. The planners and the community can only block it.

This doesn't stop their being claims that Melbourne 2030 has failed, because it hasn't facilitated the sorts of infill development that the planners envisaged, or prevented the sort of infill development (outside of activity centres) that local residents dislike (and Melbourne 2030 claims to discourage). Nor should anyone expect there to be, in the absence of any substantive changes to the planning framework to raise the costs of development in poor locations, and lower them in others.

It is somewhat specious however, to claim, as The Age did this morning, that the impact of Melbourne 2030 in the city of Monash is nothing. Not because it isn't nothing (it may be), but because the article in question (by Peterson, Phan and Chandra, "Urban infill: extent and implications in the City of Monash.", People and Place v16,i4) does a poor job of showing that to be the case. They claim, in essence, that because only a low percentage infill development occurred in activity centers (4.65% within 400m, 20.30% to 800m) or around railway stations (7.2% to 400m, 35.7% to 800m) Melbourne 2030 is failing to concentrate development.

The government response, that they didn't expect more than the 26.1% activity centre share, up to 2005, is equally difficult to parse. The problem lies in the interpretation of expectations, of what a low figure is, and of where that development would occur anyway.

By not providing comparative figures for the percentage of residential land area captured by the 400/800m zones around activity centres and stations, Peterson et al, leave me clueless as to whether 7.2% is a significant percentage (which it might be if only 1% of all land was near a railway station), or worse than random (if around 10% of land was). Similarly, it is probably ludicrous to expect no infill development outside of activity centres, so the comparison should be the level of infill relative to different areas. By neither showing, nor even defining what level of increased activity is expected, the government leaves no basis for making that comparison, and the authors have no way of determining if Melbourne 2030 has failed.

Finally, it is reasonable to deduce that developers would prefer to be near railway stations, all things being equal, so the real question regarding the effectiveness of Melbourne 2030 is whether it has been successful at driving development towards activity centres, above and beyond the expressed preferences of developers, or, whether it has been successful at enabling infill in line with developer and planning preferences. Most likely, as the article concludes, land is in such short supply that development is being driven by 'opportunism', and the expressed preferences of Melbourne 2030 are largely irrelevant to the operation of the infill market.

But That doesn't mean Melbourne 2030 has "failed". In order to fail, someone would need to define what level of housing infill would constitute a success.

Sterner Matters 24th February, 2009 16:27:44   [#] [2 comments] 

  [next -->]