Eesti teadus 2019 arvamuslugude kogumikus näitavad Allik ja Lauk graafikut, mille kohaselt Eesti teaduse kvaliteet mõõdetuna viidete arvuga artikli kohta on viimase 20 aastaga eksponentsiaalselt kasvanud. Tore muidugi näha andmeid ja arvutusi, mis avaldatud arvamustele ka mingi aluse annavad, aga Alliku ja Laugu valitud suhtarv ei pruugi just kõige parem teaduse mõõt olla. Seda indeksit tõstab näiteks nõrgemate artiklite mitteavaldamine, isegi kui need artiklid tegelikult olulisel määral uut teadmist loovad (mis pole sugugi kindel – olen sellest kirjutanud https://sanderheinsalu.com/ajaveeb/?p=558). Näiteks kui suure teadusrahastuse korral avaldataks 2 artiklit, üks ühe ja teine kahe viitega, siis oleks viidete arv artikli kohta 1,5, aga kui väikese teadusrahastuse korral avaldataks ainult see kahe viitega artikkel, siis oleks suhtarv 2. Teaduse näiline tase võib niisiis tõusta rahastuse vähenemisega. Madalam finantseerimine paneb osad inimesed teadusest lahkuma, eeldatavasti pigem nõrgemad teadlased, kes kirjutaksid akadeemilisse maailma jäädes vähemviidatud artikleid. Samuti kui värsked doktorantuuri lõpetajad avaldavad madalamalt tsiteeritud töid kui kogenud professorid ja rahastuse alanemine vähendab noorteadlaste sissevoolu, siis suurendab see viidete arvu artikli kohta.
Teine mehhanism kuidas tõsta viidete arvu artikli kohta teaduse tegelikku taset parandamata on suunata avaldamine ümber valdkondadesse, kus keskmine viidete arv artikli kohta on suurem. Näiteks eksperimentaalfüüsikas kogub publikatsioon keskmiselt palju rohkem viiteid kui matemaatikas. Kui vähendada matemaatikute ja suurendada füüsikute arvu, aga jätta keskmise teadlase avaldatud artiklid ja saadud viited kummaski valdkonnas selle valdkonna keskmiseks, siis kogu Eesti teadlaskonna viidete arv artikli kohta tõuseb. Lihtne numbriline näide on, et riigis on 1 matemaatik ja 1 füüsik, kumbki avaldab 1 artikli aastas. Kõik matemaatikaartiklid saavad 4 viidet, aga füüsika omad 10, keskmine viidete arv artikli kohta niisiis 7. Kui asendada matemaatik teise füüsikuga, siis tõuseb viidete-artiklite suhtarv 10 peale. Seda võib saavutada teadusrahastuse matemaatikalt ära võtmise ja füüsikasse suunamisega.
Adjustable-length metal crutches click with each ground contact and lifting, which some people find annoying. The reason for the clicks is that the push button used to adjust the length of the crutch is not snug in its hole, but pushes against the top of its hole when the crutch is pressed against the ground and against the bottom of the hole when the crutch is lifted.
The push button length adjustment system consists of two pipes inside each other, with holes in the outer pipe and a spring-loaded button on the inner one. Pushing the button in allows the pipes to slide relative to each other. When the button is released and pops out into a hole, it locks the pipes together. Similar push button systems are used to adjust the handle length of rolling luggage and the weight stacks of gym equipment.
To silence the clicks, the button should push against the same side of the hole at all times, which can be achieved by adding one spring. The spring should pull the inner pipe of the adjustable part of the crutch in the direction of shortening the crutch, i.e. the same direction as ground contact. Then the button always stays in contact with the same side of its hole instead of alternately hitting the two opposing sides. The reason the spring should pull in the same direction as ground contact is that the upward force on the adjustable inner pipe when the crutch bears the weight of the user is much greater than the downward force of the weight of the inner pipe when the crutch is lifted. Thus it is easier to overcome the downward force using a spring.
A homemade version of the spring is to tie a rubber band under tension to above and below the adjustment button. A bungee cord or bicycle inner tube would work as well.
The length adjustment of the crutch with a spring would be similar to that of an office chair – automatic in one direction, but requiring force in the other. When the button is pushed, the crutch shortens automatically by one hole. To lengthen the crutch, one has to push the button in and then pull the crutch in two opposing directions.
Empirically, articles with more authors are cited more, according to Wuchty et al. (2007). The reasons may be good or bad. A good reason is that coauthored papers may have higher quality, e.g. due to division of labour increasing the efficiency of knowledge production. I propose the following bad reasons, independent of potential quality differences between coauthored and solo articles. Suppose that researchers cite the works of their friends more frequently than warranted. A given scientist is more likely to have a friend among the authors of an article with a greater number of collaborators, which increases its probability of getting a „friendly citation”.
Another reason is defensive citing, i.e. including relatively unrelated papers in the reference list before submitting to a journal, in case the referees happen to be the authors of those works. The reason for adding these unnecessary citations is the belief, warranted or not, that a referee is more likely to recommend acceptance of a paper if it cites the referee’s publications. The probability that the set of referees overlaps with the set of authors of a given prior work increases in the number of authors of that work. Thus defensive citing is more effective when targeted to collaborative instead of solo papers.
The referees may also directly ask the author to cite certain papers in the revision (I have had this experience). If the referees are more likely to request citations to their own or their coauthors’ work, then articles with more authors are again referenced more.
Valderas et al. (2007) offer some additional explanations. One is measurement error. Suppose that letters to the editor, annual reports of the learned society, its presidential inaugural addresses, and other non-research in scientific journals are counted as publications. These have both fewer authors and citations than regular research articles, which creates a positive correlation between the popularity of a piece of writing and its number of authors.
If self-citations are not excluded and researchers cite their own work more frequently than that of others, then papers with more authors get cited more.
Articles with more collaborators are presented more frequently, thus their existence is more widely known. Awareness of a work is a prerequisite of citing it, so the wider circulation of multi-author publications gives them a greater likelihood of being referenced, independent of quality.
Raamatu eelis filmi ees on, et seda saab nautida omas tempos – kes loeb kiiremini, kes aeglasemalt. Olenevalt raamatust võib ka sama inimene lugeda kas kähku üle rea või pikaldaselt üht kohta mitu korda. Midagi sarnast saab teha filmiga: kerida kas tagasi stseeni uuesti vaatamiseks või edasi mõnest kohast üle. Samuti saab vaadata kiirendusega, aga siis läheb heli kaotsi. Probleem stseenidest üle hüppamisel on kas ebatäpsus või ebamugavus: minuti kaupa kerides võib mõni huvitav koht vahele jääda, aga paari sekundi kaupa hüpates peab pidevalt klõpsama ja heli ja dialoog muutuvad raskesti jälgitavaks.
Äriidee on teha programm, mis filmi kiiremini esitab kui vaatajal igavaks läheb. See kiirem esitus peaks olema „loomulik”, ideaalis märkamatuks jääv, mitte teatud vahemiku kaupa hüppamine või video kiirendusega näitamine. Märkamatuks kiirendamiseks tuleks tuvastada filmi stseenide piirid ja pikemate stseenide lõpud ära lõigata, eriti staatiliste stseenide (armunud vaatavad üksteisele silma, vaenlased põrnitsevad või ähvardavad). Filmi sisu jääb ju arusaadavaks kui silma vaadatakse kaks sekundit kolmekümne asemel. Samuti tulistamine või kaklus võib kesta mitte minuti vaid paar hetke – filmi mõistmist see ei takista.
Staatilist stseeni peaks saama arvuti abil tuvastada, sest pildi pikslitest paljud ei muutu pikema ajavahemiku jooksul. Stseeni väljajäetavat osa saab samuti automaatselt valida, näiteks kõige vähem muutuva pildiga ajavahemik, mis moodustab etteantud protsendi stseeni pikkusest. Kiiresti muutuvaid (märuli)stseene tuleb ilmselt algul inimtööga üksteisest eristada ja nende piirid filmifaili märkida. Edasi saab juba arvuti otsustada soovitava kiirendusprotsendi põhjal kui suur osa igast stseenist ära jätta.
Lisaks stseeni lühemaks lõikamisele saab märulit ja muid dünaamilisemaid juppe ka pisikese kiirendusega (edasikerimise mõistes) näidata, eriti kui neis dialoogi pole. Kui heli on oluline ja moonduks kiirendades (dialoog näiteks läheks kiledahäälseks), siis tuleb see videost eraldada ja algse kiirusega esitada, teatud juppe eemaldades, et see ajaliselt uue pildiesitusega sobiks. Lõigatavate ajavahemike valik võib jällegi nõuda inimese otsust.
Programmi edasijõudnum variant tuvastaks igavuse vaataja näoilme põhjal ise, aga algelisem võimaldaks lihtsalt kasutajal nupuvajutusega vaatamist kiirendada või aeglustada.
Interacting with colleagues is like compatibility of programs, tools or machine parts – an individually very good component may be useless if it does not fit with the rest of the machine. A potentially very productive worker who does not work with others in the company does not contribute much.
The difference between an individual and a firm may be horizontal (different cultures, all similarly good) or vertical (bad vs good quality or productivity). The horizontal compatibility with colleagues includes personal appearance – wearing a shirt with a left-wing slogan may be fine in a left-wing company, but offend people in a right-wing one, and vice versa. When colleagues take offence, the strong emotions distract them from work, so a slogan on a shirt may reduce their productivity.
Vertical fitting in includes personal hygiene, because bad breath or body odour distracts others from work. Similarly, loud phone conversations or other noise are disruptive everywhere.
Testing illegal drugs for the active ingredient differs from testing for poisonous adulterants. Both tests have opposite effects on drug use before and after buying. After the pill has been purchased, testing reduces use, because sometimes the drug fails the test, whether correctly or not, and is discarded. Before purchase, the option to test for and avoid adulterated or inactive drugs reduces the buyer’s risk, thus increases use.
In the longer term, testing benefits the dealers of purer, more predictable and less toxic drugs, putting some suppliers of fakes out of business. Pill predictability reduces overdoses – a health effect similar to lower toxicity. If old drugs can be tested, but new ones not, then buyers experiment less and the incentive to invent new narcotics decreases.
The avoidance of poisonous adulterants is good for public health, but purer pills not necessarily so. Inactive drugs undermine consumer confidence in the illegal market, reducing use, prices and casual purchases. Trust then requires a long-running relationship with the seller, which has multiple benefits. It motivates dealers to care about the health of their loyal customers, simplifies policing and gives researchers and social workers better long-term access to the at-risk population.
One claimed benefit of party drugs is that they reduce anxiety, increase the user’s confidence and social interaction, thus improving mental health. Evidence from psychiatric medicines suggests that many such benefits are due to the placebo effect. Users are quite inaccurate in estimating the purity of ingested drugs, and factors like price and place of purchase strongly influence their perception of purity. The price per pure gram is negatively related to purity in some markets, further supporting the placebo interpretation. If inactive pills boost confidence similarly to illegal drugs, then there is a clear case for flooding the market with harmless placebos. The availability of pill tests for the active ingredient reduces this opportunity to make the illegal market inefficient. Tests for toxic adulterants, however, actually favour harmless placebos.
Shoes are typically of thinner material in the toes than around the ankle, but human toes are more cold-sensitive than ankles, because toes have a greater ratio of surface area to volume (thus greater heat loss) and are further from the core of the body, so get less warm blood supply. Similarly, pants are usually thicker on the butt (due to pockets) or the upper end in general (suit pants lined to the knee), in spite of the legs requiring more warming than the pelvic region. Suit jackets are open on the chest, but overlap on the belly, which needs less extra insulation. The same suboptimal distribution of warmth characterises various V-necked upper body clothes. Jackets are also thicker on the front than the back, despite most people’s backs being more cold-sensitive than bellies.
This impractical design can be explained for men’s jackets by the desire to improve the wearer’s looks with the visual illusion of broad shoulders created by the V-shape of the front of the jacket. I have written about this in more detail: https://sanderheinsalu.com/ajaveeb/?p=885
For other clothes and shoes, there seems to be no reason for the suboptimal distribution of thickness, which would be easy to fix in the manufacturing process. The extra layers of cloth added by pockets could be balanced by using thinner cloth to make the pocket area of the pants. The pocket pouches on jeans for example are usually of thin cloth, which is a step in the right direction. However, the surface of the pants covering the pockets is usually of the same cloth as the legs. Lining could easily be added to trouser legs to make these as thick as the upper part, compensating for the extra layer of cloth by manufacturing the trousers out of thinner material overall. Similarly, adding a layer to the back of a jacket is easy.
The only difficult part to compensate in the impractical thickness distribution of clothing is the thin chest cover (relative to belly and back) of a V-neck jacket, but this difficulty only arises from the desire to preserve the look of a V-neck. A similar visual illusion of broad shoulders could be created by painting a V-shaped pattern on the garment.
A different impractical aspect of shoe design that can be explained by fashion is the pointy toes. The tapering tips create an optical illusion that makes the feet seem longer, which does not necessarily improve the wearer’s looks. However, fashion is frequently ugly, as evidenced by the web search results to the phrase: „it’s called fashion, look it up”.
Somewhat counterintuitively, moving a part of the body a greater distance may be easier in some cases. For example, lying on your back and lifting straight legs off the floor, the muscles work harder when the legs are close to the floor than when they are close to vertical. Leg lifts lying on your back are easier when their amplitude is larger (90 degrees as opposed to 45 degrees off the ground).
In many exercises, lifting the limb to an easier position gives the muscles a rest, making the workout less intense (calories burned per unit of time) overall. Examples are biceps curls until the forearm is vertical, straight arm raises all the way overhead, deadlifts to a straight or even backward tilting posture, as opposed to stopping partway through. Slower movement may make an exercise more intense by spending more time in an effortful position, e.g. slow push-ups or squats.
Lifting a longer distance may also make an exercise easier by giving a greater opportunity to swing the limb and use inertia, which is usually bad technique. For example, standing leg lifts to the front take less effort when the leg starts from behind the body and is already moving when passing vertical, compared to starting from holding the leg slightly to the front of the body.
Many low-skill jobs (guard, driver, janitor, manual labourer) permit on-the-job consumption of forms of entertainment (listening to music or news, phoning friends) that became much cheaper and more available with the introduction of new electronic devices (first small radios, then TVs, then cellphones, smartphones). Such entertainment does not reduce productivity at the abovementioned jobs much, which is why it is allowed. On the other hand, many high-skill jobs (planning, communicating, performing surgery) are difficult to combine with any entertainment, because the distraction would decrease productivity significantly. The utility of low-skill work thus increased relatively more than that of skilled jobs when electronics spread and cheapened. The higher utility made low-skill jobs relatively more attractive, so the supply of labour at these increased relatively more. This supply rise reduced the pay relative to high-skill jobs, which increased income inequality. Another way to describe this mechanism is that as the disutility of low-skill jobs fell, so did the real wage required to compensate people for this disutility.
An empirically testable implication of this theory is that jobs of any skill level that do not allow on-the-job entertainment should have seen salaries increase more than comparable jobs which can be combined with listening to music or with personal phone calls. For example, a janitor cleaning an empty building can make personal calls, but a cleaner of a mall (or other public venue) during business hours may be more restricted. Both can listen to music on their headphones, so the salaries should not have diverged when small cassette players went mainstream, but should have diverged when cellphones with headsets became cheap. Similarly, a trucker or nightwatchman has more entertainment options than a taxi driver or mall security guard, because the latter do not want to annoy customers with personal calls or loud music. A call centre operator is more restricted from audiovisual entertainment than a receptionist.
According to the above theory, the introduction of radios and cellphones should have increased the wage inequality between areas with good and bad reception, for example between remote rural and urban regions, or between underground and aboveground mining. On the other hand, the introduction of recorded music should not have increased these inequalities as much, because the availability of records is more similar across regions than radio or phone coverage.
To prevent meetings from running over because some people like to listen to their own voice, one way is to publish how much of others’ time each participant took. Measuring the talking time and making the results public helps participants with low self-awareness realise how long they talked, and creates social disapproval of those who go on for too long, potentially motivating them to be more concise.
A related method to prevent time overruns using current meeting rules, e.g. Robert’s Rules, is to allocate each speaker a fixed amount of time in advance. The problem with this method is the lax enforcement both during and after the meeting. If a speaker goes over and does not respond to requests to stop, then the moderator or chairperson usually does not shut the speaker up (turn off the microphone, forcefully remove the waffler from the stage, clamp a hand over their mouth). After the meeting, the possible sanctions (e.g. not inviting the speaker to future meetings, monetary fine, opposing the speaker’s proposed policy) are also infrequent or weak. Of course this enforcement problem also arises when talk time is recorded and published. However, the clear measurement removes one excuse of the speakers going over, namely their flat denial that they took more time than allocated, or more than others.
Public time-recording is especially helpful in less formal meetings that have no moderator or chairperson keeping time and notifying speakers to stop, and in meetings where a speaker is powerful enough that other participants are reluctant to interrupt with reminders of the time limit. A timekeeper is not needed to record the duration of a speech nowadays, because smartphones can identify a person based on their voice and calculate the time for which each voice spoke. There is a business opportunity in developing an app that identifies the number and timing of the speakers. The resulting data could also be used for research into social dynamics, e.g. whether some age, gender or race groups speak less, whether people in positions of power talk and interrupt more.
A smartphone app can also play a notification sound when a speaker’s time is up, eliminating the problem that the less powerful participants do not remind an important speaker to stop. In large meetings with a microphone, a computer keeping track of speech durations can force a speaker to stop by cutting power to the microphone when the time is up. A computer may be attached to other means to stop a speaker from unreasonably taking others’ time, e.g. it may draw the stage curtain, turn off the stage lights or start noise-cancelling the speech.