Spring wisteria: early

 I'm recording here 2020's spring flowerings, the wisteria and jasmine in particular.

This year, the jasmine started coming out mid-August.  This is roughly what I would hope for in a normal season: it was initially a late winter blossoming.  In the recent climate-disrupted past, it had been coming out early to mid July at the earliest.

The wisteria started appearing last week - around the 10th of September.  When it first started flowering, it had come out in early-mid spring, the first few weeks of October.  It hasn't come out this early before.

I have to emphasise that this represents the immediate micro-climate at the front of our house.  Other jasmine and wisteria around Sydney - and even around Coogee/Randwick - tends to flower earlier than ours, by several weeks.

So: in 2020 the jasmine came out at a more traditional time, and the wisteria is out a fair bit earlier.

If I were to pick a reason for the wisteria's differential, I might guess that because it spread onto the Bottlebrush and Lilly Pilly trees outside the gate, at least part of the wisteria is getting significantly more sun.

As you can see, it's all white this year - so far.  When I planted it, the wisteria was a light side of purple.  After a few years, the flowers along the side path emerged white, while the rest was purple.  Now it seems to be white everywhere.  On past experience, a few purple flowers may turn up in a few weeks.

Trump's final strategy: discord and divisiveness

Trump's poll numbers have drifted down to dangerous levels, due in no small part to his lack of leadership on COVID-19.  In fact, from mid-April he had deliberately pushed full responsibility to the States (see this report from the Guardian and this timeline).

With his hallmark rally plans in tatters due to COVID and spoilers, I estimate his ultimate strategy to win re-election is to fall back on pure divisiveness, in a two-pronged approach:

External bellicosity: ramp up the discord with China.  I see the forced shutdown of China's Houston consulate as part of this strategy.  It may well be true that there espionage activities have emanated from there, but the White House deliberately chose to take action at this particular juncture.  Trump may have hoped China's retaliation would go as far as closure of the US consulate in Honk Kong, but China's leaders are clearly going to strategise well above Trump's level

Internal friction: fan the flames of chaos by sending federal troops to US cities, ostensibly to quell violence.  My friend Bill in Portland told me protesters there "made a mistake attacking the federal building" - that's what gave the White House an excuse to move in to Portland; what followed was not just good tabloid fodder, but also summary policing, including unidentifiable agents beating protestors and arresting them without charge.  Four days ago, Trump intended to send them to Chicago too, but in the face of resistance from Chicago's mayor, it hasn't happened - yet.  Several news services report it as being "unclear" what the federal officers would do there - the biggest shooting reported there was intergang violence at a funeral (no deaths), which is a State matter in the first instance.

But chaos is good for business, when you're someone like Trump.  Expect more before November.

COVID: NSW spreads inside (Shape of an emerging outbreak, part 2)

The original outbreaks in Sydney eventually died down, as specific public warnings, testing, tracing, monitoring, restrictions and quarantining incoming travellers slowly did the job.

But Sydney remains on the cusp of a renewed outbreak due to infection leakage from Melbourne, and it's still too hard for me to get a sense of whether new infections will settle down or not.  The recent Victoria mass outbreak breeds pessimism; yet the non-explosion in new infections plus the record amount of testing (see bottom of this article) gives room for optimism.

Over the past week or so, with intensive testing and contact tracing, the shape of the clusters is known.  New infections still hover around ten per day, with nearly all infections traceable to an existing cluster.

Despite the graphic given in the previous post, the actual defined clusters are slightly more nuanced.  If you read the news daily, you learn the immediate source of the infection, the source behind that (which is apparently what is represented in that graphic), and the warnings of locations those newly infected have attended and thus who should now get tested/isolate.

At this early stage, it would seem that nearly all new infections are due to prolonged exposure indoors.  That is, more than a few minutes spend indoors in the same space as someone who is infected.  That could mean a pub, a restaurant, a workplace, a gym, a church.  That's it, so far.  The implication, which is understandably not reported, is that outdoors exposure, brief exposure, public transport and shopping centre exposure, is not currently causing infections.


But it's early days in this second outbreak, and that could change.  Complacency is dangerous.  I warn my late teenagers to be cautious at shops, bus stops, public transport.  Masks - of any sort - fully recommended.

A major reason this virus has been so hard - and a good reason for all the above precautions - is that an estimated 40% of infectious people are asymptomatic.  Did you attend that funeral or this restaurant?  Then the official line is that you are urged to get tested and self-isolate. 

COVID: The shape of an emerging outbreak?

The current situation:

Australia nearly had the virus fully under control, with little to no community spreading left.  Until.

Unfortunately, returning travellers were isolated for 14 days in hotels.  Not a problem in itself, but in Melbourne hotel-based enforcement was contracted out to major security companies, who subcontracted it to subcontractors, etc, resulting in non-professionals being recruited by WhatsApp, told little about the gig, and put on at short notice with scant training.  Everything went to pot from there.

Melbourne is now trying to tamp down a full-fledged outbreak, where new infections are hovering around 400 per day.  Despite the border closure between Victoria and New South Wales, infections have leaked through to Sydney, mainly via the road transport industry and Sydney pubs.  Yes, the border is a little too porous, and Sydney social distancing enforcement too lax.  Sydney infections today: 19.  We do contact tracing, and nearly all those 19 are traceable to one of the clusters.  So far, with a handful still under investigation.


In that context, ABC (the Australian Broadcasting Corporation) has in the middle of their daily coronavirus briefing included an interesting graphic of the cluster origins:

It's not easy to fathom the cause of the differential in cluster origins.  Melbourne's pattern may be due to a more mature spread pattern, and Sydney's pattern's possibly reflect the way the outbreak is emerging (Sydney's clusters include a gym across the road from one of the pubs).

But I haven't been able to fathom why Melbourne's clusters don't include pubs and restaurants.  The regulatory environment seems to be rather similar, and Melbourne people are by all accounts behaving as incautiously as those in Sydney.  Pubs in particular are great disinhibitors, so why is it not so in Melbourne?  Answers, please on a postcard, letter or emails.

I recommend the above ABC daily briefing.  Meaningful insights often emerge.

Update 28-Jul-2020
Five days later, the cluster picture is not significantly different, apart from some relabelling:

For Victoria, Retail remains a minor source, while pubs and restaurants still don't figure - that will not change, since they are now in lockdown.  The Hotel source in Victoria was the location at which overseas returnees were isolated, but contamination spread via security guards.
It must be stressed that this represents identified clusters, and doesn't include single or unidentified sources.  However, Victoria has hundreds of cases per day dating back a full three weeks whose source is still labelled "under investigation".  NSW is still hovering around 10 - 20 new infections per day, most of which are traced to source.  Because our tracers currently have a manageable load.

COVID: for too many people, it's too hard

From the ABC today:

"Biosecurity expert Professor Raina MacIntyre said she was still noticing a worrying level of complacency in NSW despite the recent increase in cases."I have noticed it myself that nobody's keeping a physical distance, nobody's wearing a mask, and unless we do things, we may lose the freedoms that we have at the moment in Sydney," she said.
"I think people should think about it as a social contract, that if we want to retain those freedoms to do the things we want to do, then in return we need to also do the right thing, maintain that physical distance of 1.5m, download the COVIDSafe app and wear a mask."

The New South Wales government is doing a precarious balancing act.  They are saying that people will react poorly to efforts to tell them what to do (I'm paraphrasing), and that it's best to encourage the right thing.

This explains why mandating and enforcement has been particularly weak.  I've seen especially egregious flouting of laws: the Royal Randwick Hotel on the day of the Big Fight last weekend allowed the punters to pack in tightly in the sports bar; staff were equivocal, and said to me they relied on management to enforce - which is fair.  Why put it on the workers?

But pubs are the worst things to keep open.  As the SMH said, alcohol weakens resolve.  And people have to talk louder when it gets noisy and crowded - so spit is more vociferous.

The long goodbye is the big hangover.  But so is the economic depression, for years to come.

New Zealand locked down hard.  And early enough.

COVID-19: Australia turning the corner

Australia's new COVID cases are plummeting:

... and the cumulative number of cases is flattening, per this logarithmic scale:

These charts and others are courtesy of the state-owned Australian Broadcasting Corporation; they are updated daily.

We're in lockdown, with non-essential business closed, and people working from home where possible.  But in so many ways, the whole word is in a strange place right now - a singularity for which we may never have a similar experience.  The solutions are well known: test, test, test, trace known cases, and isolate.  But testing depends on the number and quality of test kits, which is quite variable around the world - as is both political and social will.

Per the top 30 countries by GDP, our death rate - the number that matters the most - is 20th, one better than China, which implemented faster and more draconian lockdowns.  the 10 worst death rates per million are:

	Country	Deaths/ 1M pop
1	Spain	368
2	Italy	329
3	Belgium	311
4	France	221
5	Netherlands	160
6	UK	156
7	Switzerland	128
8	Sweden	89
9	USA	67
10	Iran	53
	:	:
20	Australia	2

However, I have learnt from reading between the lines on the numbers is that comparisons are fraught.  Australia, for example, shares no borders with other countries, which is a great advantage.  New Zealand: even more isolated.  Numbers of cases and deaths reported is quite variable, depending on the country and how much testing they do.  Sweden is another exception, as they did not lock down, which is something they have come to regret.

My friend Bill says you should look at the US by individual State.  That puts his State, Oregon, way down with only one death per million.  New York is at top, with 32 deaths per million, followed by New Jersey with 17.  Also above Australia and China's 2 deaths per million is: Louisiana, Michigan, Conneticut, then 11 more.


Actual numbers will never be certain.  I expect that a few years after it's over, analysts will get closer to true numbers by interpolating from existing death rates in the previous few years - and we will be surprised to find the countries whose COVID-related deaths are much higher than reported, as not all COVID deaths are known cases.  This will clearly be so for countries like Indonesia, India and Iran, but may also surprise us in countries like USA, China and UK.

I expect that at least in part, the eventual numbers per country will depend on social cohesion, and willingness to heed lockdowns and isolations.  Governments can mandate or recommend, but it's up to the people as a whole whether they do the right thing.  Complacency is literally a killer.  I see it well in evidence in Sydney - although we have done well by and large by clamping down reasonably early.

How will the world look when we come out of this?  I don't know what permanent changes will be wrought, but I'm pretty sure it will take several years for economies to recover, and many already-disadvantaged people will be far, far worse off.  Expect to see a longer tail of mortality from other poverty-related causes, particularly in less developed economies and pockets of the US.

A few insights into Blog comments spam

People have been spamming blog comments for years.  From manual beginnings, people developed spambots to automate the process, which understandably resulted in a huge increase in spam traffic.  In earlier days, the intention was a mixture of attempts to build traffic to legitimate sites (for both manual traffic and search engine optimisation),  and various scams including pump and dump.

Many blogs responded by turning off comments altogether; in my case I've vetted comments before publishing.

In more recent years, the spam comments have become apparently innocuous, and don't even try to link to other web sites.  This left me wondering what was going on, but I found it difficult to find out.

So with my previous post, I set up a honeypot to gather spam comments over time, with the intention of analysing them to understand better what was going on.  This entry will give some limited insights; maybe I will add to it when I know more.  The following is a first pass report on the results.

I posted the honeypot in August 2019, and spambot comments flowed in for about two and a half months, before abruptly stopping.  There's an implication that they're all from the same source - or using the same mechanism.

I analysed about 100 comments through a Natural Language Processing framework.  This is a form of Machine Learning (which is popularly referred to Artificial Intelligence, although I don't think it's an accurate term).  It wasn't able to tell me that much.  Amongst other things, it returned a high positive sentiment score through sentiment analysis.  This was fairly obvious already.  To get past spam vetting systems, the spambots intentionally fed relatively upbeat phrases.  They were mostly quite general comments, either about liking the blog or asking help with their own blog (again, no  links).  But it was possible to tell in the first instance that it was spam simply because there was no specific reference to the subject matter in the blog post.  To make this clear, in the honeypot I requested comments to include a specified word to flag that the commenter had read the post.  Which of course is beyond the capability of automated commenting tools.

The only thing I've really gotten from the NLP system so far is that very frequently the comments are very close variants on each other - in groups of two, three or more.  It's as if someone put together three sentences, made some variants on a few keywords/phrases, and then got the spambot to switch around the words frequently enough so as to specifically avoid getting caught by automated processes that blocked groups of indentical comments.

So it looks like it's an arms race between sets of automated tools, a battle to infuse comments on the one hand, and to deflect them on the other.  What hasn't been answered yet to my satisfaction is why the spambots are still running but are not delivering weblink payloads.  My only guess is as before: that the spambots are being used to pinpoint blogs/news sites that allow unfiltered comments to get through.  My feeling is that there must be more to it than that, so suggestions are welcome.

PS: Will a new post get those spambots started again on this blog?  Let's see.