Older blog entries for chalst (starting at number 197)

Recentlog

• Recentlog is quite messed up recently; for example: indentation following <br> tags is typeset rather strangely for several diaries right now, mod_virgule is resisting my attempts to set diary ratings for nuisances such as testor (as well as for underrated newcomers such as beraldo), ...
• marnanel writes about Things I never expected to hear a politician say, so reminding me of Carl Bildt, the centre-right Swedish Foreign minister, who was unusually blunt in his recent interview with the FT: South Ossetian independence is a joke. We are talking about a smugglers’ paradise of 60,000 people financed by the Russian security services. No one can seriously consider that as an independent state.

Grokking Finance
Last week, apenwarr linked to Havoc Pennington's elementary exposition of a useful tool of financial analysis, Return on Equity, part of Havoc's promising looking series on how understanding business concepts can be useful for folks trying to organise things together.

At type other end of the scale, to try to understand the limitations of business and finance techniques for non-specialists, maybe it is worth trying to understand why these fields seem to behave strangely. I can't claim to really understand the pathologies of MBA-driven thought, but I did read an excellent opinion piece on why finance seems to be attracted to reckless gambles, Frank Portnoy's FT opinion piece from Monday, A crisis of similar ingredients.

Portnoy's analysis suggests three common factors behind financial crises that he observed in the last two major crises, from 1994 and over the past year, and that he suggests these are necessary features of financial markets:

1. Informational asymmetry: which arises from the knowledge gaps in how financiers make sense of data and in modelling risks. For example, he notes that in 1994 when Askin Capital Management applied mark-to-market accounting to its assets, it suddenly reported that it was insolvent, "the financial markets were stunned". and goes on to note that "not very many people can price complex financial risks accurately, especially those related to mortgages".
2. Moral hazard: he emphasises moral hazards arising from unforseen consequences of guarantees made by government sponsored enterprises, but he might also have mentioned the moral hazards arising from accountancy standards, or the idea that some firms are too big to fail.
3. Regulatory arbitrage, "the manipulation of legal rules for financial advantage". He observes that the special legal status of AAA-rated instruments have resulted in credit rating agencies being at the centre of both the 1994 and the current crisis. He notes that "highly rated structured finance instruments are incomprehensible to most investors", and of course, the risk implications of highly complex structured instruments are not well understood by the ratings agencies.

Portnoy ends, noting that "Information asymmetry, moral hazard and regulatory arbitrage are the basic ingredients of high-margin finance. Soon the markets will be booming again, and the people who exploit these three themes will do the best, as they always have".

Mode 7
tagishandy links to some 8x8 BBC micro fonts he has made available. I saw that acb had made available some BBC teletext fonts. I think those are not the exact same fonts as per the BBC micro teletext chip, though they are of BBC provenance.

The menace of new accounts
lkcl writes some advice on creating new accounts that would, if followed, save legitimate accounts from being deleted, and certs jbaker up to apprentice.

ncm summarises the spammer menace:

My experience is that the account deletion threshold is set so (artificially) high that the only way to get a spammer account deleted is to mark them down as spammers unless it's dead clear they're not. Even then lots of fake accounts get through. Mostly the real spammers don't start spamming immediately; they appear to be accumulating fake accounts and holding them in reserve.

Well, fair enough, this shoot first, ask questions later approach will certainly reduce the conceivable possibilities for spammers to abuse Advogato, but it's not how the spam rating feature was meant to be used, and the history of Advogato has been to apply fixes, rarely needed, when problems actually emerge.

You're advocating aggressively marking accounts as spam without articulating a clear threat to Advogato: what are spammers going to do with their arsenal of cert-juiceless accounts? I'm bothered that it makes Advogato less friendly to neophyte developers without serving any useful purpose.

Overzealous culling of new Advogato accounts
I see that the new account jbaker, linking to http://bakerdesignz.com has attracted a spam rating of 13. Looking at the articles posted there, this seems to be the business site of a Justin Baker who uses javascript in his work, and wants to share some site design and javascript under GPL. This looks worthy of an apprentice certification, and certainly should not be counted spam.

In general, as Steven Rainwater has already pointed out, a lot of new accounts that don't spam anything are being marked as spam by quite a few people. Unpromising new accounts don't weigh much; don't mark as spam accounts that aren't actually being abusive in some way.

Monotone vs. git
Observe: Reconsidering the work flow and how the SCM system fits in on the Openembedded-devel mailing list.

Questions:

1. The motivation for the move was difficulties they had with mtn's propagate. What, I wonder, were they were doing different with their repositories than other users?
2. Mamona migrated their repository to git. How does this work? Are there conversion tools? Did they agree upon a time when all of the developers converted their repositories? Or did they have encroaching gradualism, where a gatekeeper converted, and others converted when they needed to talk to the gatekeeper?

Postscript: I guess tailor can handle the project migration.

Food prices: stabilisation?
The Financial Times observes that the UN FAO food prices index has stabilised, showing a tiny decrease for April (216.7, compared to the revised 217.0 for March). I'm guessing that this is a response to the high predicted wheat harvests for this year. This is, by no means, the end to the food price crisis, but it might end the recent supply panic.

Martin Wolf's column for two weeks ago, Food crisis is a chance to reform global agriculture, argued that speculation is unlikely to have played much role in the ramping up of food prices, since food price inventories have been so low. So we shouldn't expect any sort of speculative unwinding of food prices: they are unlikely to "bounce down" in the way that stock markets have done. Instead the reason for high food prices has been rapidly growing supply not keeping up with even more rapidly growing demand (part of which is the growing demand for meat, particularly in China) and cost of inputs to agriculture, particularly oil.

Paul Collier's comment on Martin Wolf's article argues that a general acceptance of GM foods, particularly in Europe, and promotion of large-scale agriculture (ie. industrialised), particularly in Africa, is an important part of any effective response to high food prices. Just threw that in, in case this post was lacking in controversial assertions.

Food prices: the ideological assault on food security
Daniel Davies asks What's that got to do with the price of wheat, rice, maize and ethanol?, talking about the dismantling of food security infrastructure in the third world, carried out in the name of free market dogmatism. Quote:

I think the underlying idea, in as much as there was one, was that international aid was a more efficient way of providing food security than domestic reserves and price controls. Which has a certain plausibility to it, as long as you only look at one country at a time and assume that food shortages will be caused by ecological famines, which are more or less uncorrelated between regions, rather than a global inflation in food prices which overwhelms the capacity of the food aid industry, and which arrives at a time of fiscal strain in donor nations. Of course, the general approach of assuming that one's risks are uncorrelated and manageable is one that has been causing all sorts of problems in the world economy of late.

Some trust-metric references

All the references listed below link to PDFs, with my comments in italics.

• Weeks, 2001. Understanding Trust Management
  Systems. Crucial paper for understanding much recent work on
  trust-metrics. Presents framework, motivates it with examples, and shows
  PKI can
  be modelled using it.
• Carbone, Nielsen and Sassone, 2003. A Formal Model
  for Trust in Dynamic Networks.. Proposes a simple domain-
  theoretic formalisation of what a model of trust is that provides the basis for
  some significant later work.
• Twigg and Dimmock, 2003. Attack-Resistance of Computational
  Trust Models. Synthetic exposition of several trust-metrics that
  have some degree of attack resistance. Good exposition of Raph's metric,
  gives alternate proof of attack resistance by order-theoretic reasoning.
• Moreton and Twigg, 2003. Trading in Trust, Tokens, and Stamps. Proposes system
  for
  adding trust trading to trust-metrics by stamps, and argues for advantages
  of
  this system. Interesting for Advogato: could provide a more principled
  means of bringing in new members than just rejigging weights.
  
• Ziegler and Lausen, 2005. Propagation Models for Trust
  and Distrust in Social Networks. Motivates criteria for success of
  trust-metrics when applied to social networks, including attack resistance.
  Proposes a new algorithm, which they call Appleseed and which is
  based on the spreading activation technique, which they
  compare to Raph's version. This work is based in part on Ziegler's PhD thesis,
  Towards
  Decentralized Recommender Systems.
  

I'm curious as to the origin of Raph's talk of "good", "bad", and "confused"
nodes. Is it his novel usage, or did he get it from elsewhere?

I'm amused also to read references sections that cite
Raph's
abandoned
trust -metrics thesis as if it was a successfully defended one. Folks who run
into
the thesis he did write will no doubt conclude he is one of the select few who
wrote two PhDs concurrently...

Food prices
I've been thinking and worrying about food prices more and more over the last four weeks, and I see that skvidal shares my worries.

In fact, he doesn't worry enough! When he talks of food prices "spiking", there's more to the story. The S&P-GSCI agricultural commodity price index, which is the main aggregate food price index, has shown a nearly three fold increase from Jan 2005 until the beginning of this year. Given the terrible access to credit poor farmers face, this is likely to mean that food prices, already difficult for many to afford, will get higher as the year progresses. If there is nothing done to relieve the situation, it will spell immense suffering and political instability in much of the world.

It's a scary picture, and if it isn't ignored by the world media, not enough attention is drawn to just how serious this all is. It's pretty damn clear that the story matters more than the bloody farce in Iraq (which has contributed to it by pushing up oil prices), and it is more important than the financial turmoil shaking the world right now (though that will make it harder to do anything to help).

Can we do anything to help? Skvidal has suggestions: I wonder if growing food is an option for me? lkcl's article Singularity of computing, might have something to do with food price stability in the future, though I don't suppose this kind of thinking about technology can help with this year's crisis.