Ready.
Ersoy and Panwar (1993) propose casting their lousy, "optimizing" hacks to a network:
"In a given network, if the underlying physical topology can be pruned to the minimum-delay spanning tree, bridges can be set up to choose this tree as the active logical topology by adjusting the parameters of the self-configuration algorithm [15]." (EP93, p.1173)
Pruning?!? Both tedious, and makes the network less resilient. Such riskiness is next-of-kin of sabotage.
If some (non-edge) LAN fails, all of the network is split -- into two or more isolated segments.
In contrast, if the network is with spare paths/bridges , the topology could modify itself (Pe85).
The transparent-bridge standard (Pe85) itself, is allowing the flexibility of the primary- vs. secondary-bridges being through alternate LANs -- a flexibility that facilitates safety (network resilience) -- while EP93 might (at most, if at all) accept some bridge redundancy (between LANs), which would provide some safety (only) against bridge failures, but nothing against (cable or node) failures within LANs.
Ironically, that "[15]" they refer, is Pe85 [4] which proposes a transparent-bridge protocol (algorithm), because
"It is undesirable to have a network that can be brought down so easily, merely by plugging a cable into the wrong place." (Pe85, p.44 (that is, on the first page of Pe85))
Contrast Pe85's resilient, casual style, vs. the tedious process of EP93 juggling all of the bridge IDs, any time the configuration changes. You would have to re-run the simulated-annealing, and re-establish the connections (wires/bridges), manually -- with attention to bridge IDs (numbers which must be set manually (if IDs are not factory-set), or the operator/technician/"Ph.D." must sort the (factory-set numbers) bridge IDs, in the topology order, to reflect optimal depth settings).
If transparent-bridges were remotely controllable (through software), under the command of the network-central management, then they might reflect the topology to the bridges remotely (rather than walking around, and manually wiring the bridges or upgrading their firmware with new priority bytes). If thus, that tediousness would be less, and automatizable. But if thus, the causes of designing the transparent-bridges, get lost. Administrators need to monitor the bridge IDs, and that becomes a type of dynamic routing. Trivial change of style. Like going to business meeting with t-shirt, or mountain-climbing with business suits. "Thinking what is unthinkable." -- for a Ph.D.
Pe85 endorses resilience, on its first page. If EP93 forgot that point (or, misunderstood), then again, while looking into the last few pages of Pe85 (50, & 51-52) that EP93 refers to as "[15]," Pe85 offers a few opinions on why and exactly how to extend that protocol (algorithm) for constraining its portions, if/when tuning (toward optimizing). That is, the research question which EP93 takes and blunders. Perlman is sensible in telling fine-tuning tips, while EP93 is (non-contributive but) excessive in constraining all of the network. Pe85 (p.51) hints,
Pe85 (p.52) is not sounding like there is full n-to-m routes flexibility in hardware, which would allow the network management to set any topology through software [remotely]. Pe85 tunes with a priority-byte (numeric 0..255).
That allows telling depths but obviously not sufficient to arbitrate the sides at equal depths of a tree, strictly (leftmost vs. rightmost vs. middle-left vs middle-right, etc). Let alone the problem of the limit of 0..255, if truly only a byte. Presumably, Ersoy and Panwar thought, or were told that, the topology of a tree would not be fully constrainable by setting node-depths -- without pruning.
Thus, EP93 needs that "pruning," but it is unacceptable. In case you would wonder whether other people in the field of network-optimization have been similarly blunderous, then look at the paper-abstracts of the prior art which EP93 cites. You might find all of the components of EP93, but utilizing those in other problem contexts (and as far as tellable from their abstracts, those other people were not necessarily blunderous). That reminds of the blunders of the plagiarist copycat82, its cutting-&-pasting, then "components" of copycat82 becoming makruh, etc.
The "optimization" which EP93 messes about, is what most undergraduate students were supposed to be capable of writing, in a day, or two. In contrast, the mess which EP93 causes, is certainly not the most trivial puzzle I have seen. Thus, you would most likely neglect EP93 entirely.
Thus, EP93 is fitting into the category of "AI 2" (namely, "solving" a trivial problem and then listing the rest of the problem as "extension" of that, left to the "future," which never comes). That is how EP93 is -- although simulated-annealing itself (when others work with it) is a quite respectable method ("AI 1," if AI, at all).
Trivially, casually, with Pe85 style, if you might notice that two LANs talk mostly back and forth, then a safe strategy is to bind the two with a primary bridge. If there is no extension to allow priority settings (primary, secondary), then take those two to one of the edges (branch nodes) of the spanning-tree (that which has less to talk with the rest of the network, being on the edge, friend next to friend, the talk costing less). Thus, the risk of minimizing (by hard-wiring the two), would at most, risk the edge LAN's becoming isolated -- presumably, in return for a better performance for that LAN. Not risking all of the network, nor becoming a traffic-bottleneck.
"It might be desirable to influence the topology that is computed by the spanning tree algorithm. One facility that might be desirable is the ability to designate some bridges as "primary" bridges, and others as "secondary" bridges. The algorithm should compute a topology in which no secondary bridges appear, unless no topology exists consisting solely of primary bridges. Also, it might be desirable to minimize through-traffic on some LANs, for instance those that might be lower speed or more congested with local traffic. Also, it might be desirable to configure the network so that topological changes will have minimal impact." (Pe85, p.51)
Brute-force simulated-annealing might find those, too. But probably, the points worth to find, will be only such common-sense tunings, that we might guess, as well as optimize for safety, too. The rest of the fixations of "all-nodes-minimizing" (unjustifiable risks), might hurt without helping.
Ironically, simulated-annealing gets applause mainly for its non-greedy strategy, while EP93 is plugging it into another sort of greedy context, which causes the trouble. In turn, the root of that greed of (lousily) "optimizing," is quite guessably the greed toward getting a Ph.D. title, rather than truly inventing a valuable method. Such mimicry (imitation) of researcherness, to get a Ph.D. title, is not unseen.
The Hebrew word "minim" turns out to mean "heretic." That fits how EP93 is trying to minimize the network.
minimizing hacks. Aggressively minimizing the network how EP93 (minim93) suggests, is a heretical notion. You would be risking your network need. Perhaps, that is "only" a company's lost business hours (everytime when a LAN fails), but sometimes lethal, in hospitals, reactors, etc.
I had found the paper, EP93, in 1995 (Aug.21), although mostly or entirely forgetting that it was in a pile, until 2009 (July 12). That word, "minim," got into my vocabulary, while then, at around the same time (in Aug/Sept.2009), I was investigating about sabbataism (a heretic, kabbalist-messianic, crypto-Jew sect, the reputations of which sound like fitting the antichrist-gang's profile). The judaists who knew sabbataists in Salonica, Greece, were known to call sabbataists "minim," and even opposing their return to Judaism in 1924, when judaists would remain in Salonica, but "muslims" would be sent to Turkey (of T.C). Thus, they sent all of sabbataists to T.C., which was (presumably) a loss of access with the (non-sabbataist) people they had known in Salonica -- a loss of their (human) network there. Perhaps, sabbataists found their loss in Salonica, as gain in Turkey, but in the case of institutional-networks which might fail because of hacks like that of minim93, presumably, few or none would profit from a loss of network access.
If EP93's "method" of supposedly optimizing the network would otherwise have merit (like being necessarily fast), you would perhaps list it as suggesting a kind of "living fast, dieing soon" philosophy to the networking business. But notice that, every time some LAN in that [house of cards] fails, people wait & lose their time, too.
Not to mention the potential false business decisions (with less than full information), when the network is not working perfectly (and until people notice the failure). That may sum up to loss.
Some people think that, a professor if worthless/false in the class, might be willingly committing a type of "know-how hiding." With that type of logic, the response of Ersoy to me ("engineers" vs. token-ring), might seem as sinister (perhaps, antinomian). But now you witness that, his Ph.D. "work" is similarly thoughtless. On purpose?!
Personally, I had not thought his falseness in the lecture as sinister (but other people express such opinions about some professors). The extension of that to his Ph.D., would be trivial, but again, I might think rather in terms of his incapability (basically, his thoughtlessness), rather than presuming his publishing foul, non-contributive texts, on purpose. (The case is strange, though.)
Looking at that strange case, we might list a few theories (or, black humor). For example, sabbataists live in Turkey (of T.C.), and various people [3, 5] think that sabbataists have mighty positions in T.C. [& influence how it works]. Then, might it have been their lobby? Some learned sabbataist elders (with a Y.Ö.K. lodge) perhaps thinking that, minim93 "merits" endorsement (professorship) exactly because it is wrong (with "faith" that "endorsing foul-things is good" [2])?
I have no personal acquaintance with Panwar. But, speaking of antichrist and broken networks, his native country, India, comes to mind. They have plenty of (so called) "godmen" there -- man literally thinks himself as a god, and has plenty of followers, too. One of the most known, even claims to be enlivening the dead. That all fits the antichrist concept. I'm not suggesting that antichrist would necessarily be one of those, but witnessing that lots of people find that somehow normal. Thus, the antichrist might have his niche, there. In the case of LAN, the "re-enlivening" option sounds fitting the needs of EP93. Rather than a network that casually runs, you would have to keep contact with your technician anytime something changes in the network, because the network is so strictly-tuning to the traffic intricacies, and worse, might be cut off [partially] if some critical link is broken.
Joke: Some people think that if there were no drug corporations, we would have much less need for drugs. (In extreme, to suggest that they poison us to sell drugs. Iatrogenic, etc.) In this case, the "Dr." (Ph.D.) Ersoy, is standing to cripple LANs. What next? What would be the cost of a repair visit, or for running a custom "simulated annealing" for some new configuration? Trivial job, but you see, that was able to get a Ph.D. title. Thus, might make "good" business, too?
[1] Ersoy, C., & Panwar, S.S. (Oct. 1993). "Topological Design of Interconnected LAN/MAN Networks", IEEE Journal on Selected Areas in Communications, v.11,no.8, pp.1172-82.
[2] Michaelson, J. (Jan./Feb. 2007). Heretic of the month: Shabbetai Tzvi, American Jewish Life Magazine. Retrieved through http://www.ajlmagazine.com/content/012007/heretic.html, on September 14, 2009. (@)
[3] Nassi, G. (1997). Secret muslim jews await their messiah: Shabbetai Tzvi, Retrieved from http://www.sefarad.org/publication/lm/009/nassi.html, on September 14, 2009. (@)
[4] Perlman. R. (Sept. 1985). An algorithm for distributed computation of a spanning tree in an extended LAN, in ACM SIGCOMM Computer Communication Review archive, Vol.15, No.4, pp.44-53. Retrieved through http://portal.acm.org/citation.cfm?id=319004, August 11, 2009.
[5] Pidcock, D.M., "Dajjals, Donmehs and Extra - Terrestrials", Common Sense, no.14, May 1995. Retrieved through http://www.islamicparty.com/commonsense/14donmeh.htm, on September 14, 2009. (@)